Commit Graph
218 Commits
Author SHA1 Message Date
Codeman maintainer 8a54ded78b Merge #784: a local model catalogue for every Mac memory size, and its weekly check 2026-11-08 11:10:13 +01:00
Codeman maintainer 83fb523979 fix(local-model): rollback copies follow each tag's own draft length
The previous commit counted 4 rollback copies of the recurrent state for every GGUF with a
prediction layer, the runtime's default `draft_num_predict`. A tag can set its own in its
params layer, and the runtime then passes that as `--spec-draft-n-max`, which is what
`need_n_rs_seq()` turns into rollback copies. Read on 2026-11-08:

- qwen3.6:35b-a3b-mtp-q4_k_m: draft_num_predict 2, so 2 copies (was counted as 4).
- qwen3.8:27b-mtp-q4_k_m: draft_num_predict 4.
- qwen3.8:27b-q8_0: none in its params, so the runtime's 4.

The 35B-A3B GGUF row's overhead drops by 132 MB: 0.971 instead of 0.976 on a 36 GB M3 Pro or
M3 Max, the same pick. test/local-model-catalog.test.ts now carries the drafted tokens per row
instead of a yes/no prediction layer.

Also states what was verified about llama.cpp's checkpoint default: 8 from b6800 to b8200,
32 from b8250 together with a checkpoint every 8,192 prompt tokens (common/common.h at those
tags), in place of an inferred reason. Checked in the b12111 source as well: the draft context
of a Qwen-family MTP model caches the prediction block alone (`mtp_on_hybrid_qwen`, layers at
or above n_layer()), and the main context leaves it out, so the row counts that layer once.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-08 10:38:12 +01:00
Codeman maintainer 2cd3cbd6a0 fix(local-model): count what llama-server keeps beside the K/V cache, and bound its checkpoints
The llama.cpp memory rule of the catalogue left out two allocations that the llama.cpp the
runtime bundles (b12111, the runtime's LLAMA_CPP_VERSION) makes for every GGUF row:

- Context checkpoints. For a model whose state cannot be rolled back (the recurrent state of
  the Qwen-family rows, the sliding window of North Mini Code), llama-server keeps host copies
  of that state, two or three per request, up to 32 per slot by default
  (tools/server/server-context.cpp create_checkpoint). For the 27B that is 32 x 157 MB = 5 GB,
  none of which was counted. The runtime now sets LLAMA_ARG_CTX_CHECKPOINTS=8 (read by
  llama-server from its environment like LLAMA_ARG_CACHE_RAM; 8 was llama.cpp's own default
  before it went to 32), and every llamacpp row counts 8 copies of its state.
- Multi-token prediction. The two -mtp tags and qwen3.8:27b-q8_0 carry a prediction layer
  (nextn_predict_layers 1, read from the first bytes of each GGUF; the 9B tags and North Mini
  Code have none). The runtime then always passes --spec-type draft-mtp --spec-draft-n-max 4,
  so llama.cpp keeps 4 rollback copies of the recurrent state (need_n_rs_seq(),
  llama-memory-recurrent.cpp) and the prediction layer's cache in the draft context's cache
  type, f16 by default. The rows counted one state and that layer at q8_0.

What moves (test/local-model-recommend.test.ts, "every memory size", re-pinned):

- 16 GB: limited -> none. MiMo 9B at 65,536 tokens needs 10.2 GB, 0.987 of the 10.4 GB
  estimate. With the budget the Mac mini measured (10.7 GB after the reserve) it is offered,
  0.954, tight; that test is kept. With 4 checkpoints instead of 8 it would be 0.966 on the
  estimate: this size sits on the edge, and only a run on such a Mac settles it.
- 18 GB: the same pick, 0.964 instead of 0.928.
- 36 GB: the same picks, 0.966 (27B on M4/M5 Max) and 0.976 (35B-A3B GGUF on M3).
- 48 GB on a Max chip: the 27B at 262,144 tokens is tight (0.853) instead of good.
- 32 GB with a measured budget of 0.74: North Mini Code at 131,072 tokens (was 262,144), the
  27B at 65,536 on a Max chip (was 131,072).
- Every MLX row, and everything from 64 GB up, is unchanged.

test/local-model-catalog.test.ts recomputes the new rule from the architecture table and holds
the catalogue to LLAMA_ARG_CTX_CHECKPOINTS. The catalogue header, runtime.ts, the types,
docs/local-models.md, CLAUDE.md and architecture-invariants say the same. Read in the source,
not measured.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-08 10:26:57 +01:00
Codeman maintainer bc9c3c1639 docs(local-model): two catalogue header claims made exact
The Q8 27B row counts a prediction layer its GGUF may or may not carry
(not checked, careful side), and the macOS notice on the runtime's page is
about Laguna XS 2.1; Laguna S 2.1 is set aside for its open runtime issues.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-08 09:39:48 +01:00
Codeman maintainer 97f10a1033 docs(local-model): the long-session caveat of the tested pair, and what the check proves
The 48 GB row now says the test was a short task and that MLX may keep up
to 8 GiB more in a long session, which brings the pair to the edge of the
budget. The check after the download says what it proves for each runner,
from the runtime's source: for a GGUF build both of its figures are the
buffers llama-server reports allocating, so the check covers the whole
load whatever the cache type; for an MLX build only the weights. And the
docs say where a new row's architecture goes.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-08 09:39:27 +01:00
Codeman maintainer 989fcd778a feat(local-model): a catalogue for every Mac from 8 to 512 GB, each row with its own figures
The catalogue had seven Qwen-family rows that all carried the memory
figures measured on one model (Qwen3.6 35B-A3B on the MLX runner), so a
9B or a dense 27B was judged with figures that do not describe it, two
rows had the wrong Mac download size, and every Mac below 32 GB got
"none".

Now eight rows, each with the runner a Mac really uses for its tag (a new
`runner` field: `mlx` or `llamacpp`) and memory figures of its own:

- the 35B-A3B MLX row keeps its measured figures and the tested pair;
- every llama.cpp row is computed from its model's config.json: a q8_0
  K/V cache for the whole context (what the runtime now sets), its
  recurrent or sliding-window state, 1.0 GB of runner and the 1 GiB host
  prompt cache;
- the one other MLX row (Ornith 1.5 122B) is computed the same way,
  scaled by what the measured row showed, plus the runner's 8 GiB of
  prefix snapshots.

Sizes are the registry's Mac pull totals (2026-11-08, lowercase ids
verified there). What each memory size gets: 8 GB none; 16 to 32 GB a 9B
(limited); 36 GB Qwen3.8 27B or the 35B-A3B as GGUF; 48 GB the tested
pair (or the 27B on a Max chip); 64 and 96 GB the 27B or the 35B-A3B;
128 to 512 GB Ornith 1.5 122B. Nothing but the one row was measured.

The catalogue test recomputes every computed figure from a table of
architectures and lets two rows share figures only when they share the
architecture and the runner. The rule tests of recommend() now run on a
frozen copy of the 2026-11-06 catalogue, so a weekly catalogue update
changes only the per-memory table and the real-Mac tests.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-08 09:33:48 +01:00
Codeman maintainer 54c02990ba Revert the skill default flip on this branch: it ships in its own pull request
This reverts commit 0aa34da8, which made the codeman agent skill default on here as well. That change lives in feat/agent-skill-default-on, together with the self-ignoring copy and the reference-folder symlink check it needs.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-08 09:00:59 +01:00
Codeman maintainer 0aa34da80f feat(settings): the codeman agent skill is on by default
agentSkillEnabled now reads an absent key as ON, the way
workspaceHooksEnabled does: only an explicit false turns it off. A new
session therefore finds the codeman skill in its folder without a
setting first, and with it the verbs that hand work to other CLIs and to
the local model on this Mac. Settings shows the switch on for the same
absent key.

The injection itself is unchanged: add-only, marker-owned, never through
a symlink, never for remote sessions or a plain shell. Turning the
setting off still removes nothing already written.

The quick-start test now pins the new default (injected on an absent
key, nothing after an explicit false), and the docs that said "default
off" say when that changed.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-08 08:13:51 +01:00
Codeman maintainer e8a63ede00 Merge feat/local-model-recommendation: recommend and install a local coding model on this Mac (experimental)
Conflicts resolved as on test/desktop-macbook: config/test-suites.ts keeps both browser tests (desktop-connect-page and local-model-ui), and the CLAUDE.md state-files line keeps #778's cli-installed entry with the local model's local-model.json and local-model/ entries.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-08 04:06:46 +01:00
Codeman maintainer 492df776d0 feat(local-model): judge the fit on the measured peak and rank a tested Mac first
The catalogue rows carry the figures measured on a Mac (the MLX build of
qwen3.6:35b-a3b is 23,105,798,956 bytes, about 37 KB held per context token,
about 3 GB of runner overhead, about 5 GB more for a moment while a long
prompt is processed). A model's need is now its peak. Rows that were not
measured use the same per-token and overhead figures and say `estimated`.

qwen3.6:35b-a3b gets the tested entry for an Apple M5 Pro with 48 GB and
OpenCode, with its measured 67.7 tokens per second, so that row may say
"Tested on a Mac like this".

The ranking changes with it: a pair tested on a Mac like this one comes
first, then speed (ok, unknown, slow), then model rank, harness rank and the
larger context. The fit grade decides whether a row is listed and what it
says, it no longer sorts, so a model is offered at the largest context it
fits at. recommend() takes the budget the runtime measured
(`budgetSource: measured`), and modelFitsBudget() judges one model against it.

Candidate rows now always carry `needSource`, `harnesses`, `installBytes`,
`installGb` and, where one applies, `tag`. `kvBytesPerToken` and the prompt
cache tiers are gone.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
2026-11-06 22:16:37 +01:00
Michael GrundbergandClaude Opus 5.5 8d33f53fd0 fix(files): harden the outside-worktree root after review
Keep the 1.49.0 API meaning: liveWorktree and root=live stay inside the
session folder, and an outside worktree is published as outsideWorktree
and walked only with root=live-any.

Refuse a symlinked .git folder as the main worktree, sensitive files
inside a followed worktree, and any outside worktree while attachment
confinement is on or around a blocked tree. The file routes re-check
the guard settings on every request, and a throwing rule check refuses.
The panel keeps its Back to worktree button while pinned to the session
folder.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-06 20:33:14 +01:00
Codeman maintainer 92c20fe572 feat(local-model): hardware probe, model catalogue and the pure recommendation
The core of the read-only recommendation: what this Mac is, what the catalogue
knows, and which model fits. Nothing is downloaded, installed or written.

- hardware.ts: one bounded async spawn of /usr/sbin/sysctl with seven keys,
  without -n, parsed as "key: value" lines by the pure parseSysctl(). Free disk
  of the data dir through statfs behind the bounded path probe. Fallbacks from
  node:os, a 60 s cache, never throws, never system_profiler. A wired limit of
  0 means "not set".
- local-model-catalog.ts: seven models, three harness rows (pi gated) and the
  chips table as a compiled constant, validated by a Zod schema on first use.
  No override file. Model ids are name:tag in lowercase without a slash; every
  model carries paramsB and an (empty) tested list. Chip names match on the
  end of the chip string, so 'M4' is not the chip of 'Apple M4 Pro'.
- recommend.ts: device budget (the wired limit when above 0, else 2/3 of
  memory below 36 GB and 3/4 from 36 GB, less the reserve), need = weights +
  KV cache + 1 GiB + prompt cache, fit good up to 0.85 and tight up to 0.98,
  speed ok, unknown or slow (a chip in no chips row is unknown), and the
  ranking. Verdicts ok, limited (best model under 12 billion parameters) and
  none with its reason. "Recommended" only for a pick with a tested entry for
  this Mac, which no row has today.
- types/local-model.ts: comments brought in line with the code (chip matching,
  budgetShare rounding, a budget of 0 on an Intel Mac).
- cli-registry-no-id-branching.test.ts: the guard also catches a comparison of
  a `cli` field with a CLI id (plus one stray blank line Prettier refused).

Tests: test/local-model-format.test.ts, test/local-model-hardware.test.ts
(the captured output of two real Macs), test/local-model-catalog.test.ts,
test/local-model-recommend.test.ts.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
2026-11-06 16:27:33 +01:00
Codeman maintainer 73ce6a1819 feat(local-model): types and module skeleton for the hardware-aware recommendation
The contract three builders code against in parallel. Nothing is recommended
yet: the route answers 501 and the catalogue is empty.

- src/types/local-model.ts: HardwareFacts, the catalogue rows, HarnessFact,
  Candidate, Recommendation, LocalModelStatus, and LOCAL_MODEL_LABELS, the one
  list of sentences the server may put into the payload (the zh-CN dictionary
  is keyed by them).
- src/local-model/format.ts: formatGb, the one place a byte count becomes a
  size a person reads (decimal GB), and memClassGb.
- src/config/local-model-catalog.ts: the Zod schema (model ids are name:tag,
  lowercase, no slash) and an empty stock catalogue. No override file.
- Skeletons with their final signatures: hardware.ts, recommend.ts,
  harness-facts.ts, routes/local-model-routes.ts (injectable dependencies),
  public/local-model-ui.js with its script tag.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
2026-11-06 14:56:13 +01:00
Codeman maintainer 800b3f4658 fix: keep foreign probe parsing linear in what it reads (release review)
The parser and classifier run synchronously on the event loop over text a
pane, a container or an ssh host wrote, and one process could stall the
server for every session: 4 s for a 128 KiB argv of `/pi/` tokens, about a
minute for 2 MiB of socket rows.

The `cli.js` launcher rule leaves the signature regex for a per-word scan,
socket rows dedupe through a Set, the process row regex takes the `s` flag
(a bare CR made `$` fail and `\s+` give its spaces back, quadratic in a row),
argv and pane command are cut at FOREIGN_MAX_ARGV_CHARS, and one matcher per
scan answers each distinct argv once, since every pane of a hostile host can
claim the same process tree.

Measured on a private tmux server with the same 131 KiB argv: the largest
event-loop gap went from 3991 ms to 45 ms. The worst scan the caps allow
takes about 80 ms. Socket rows stay uncapped on purpose (see the docs).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-06 07:10:49 +01:00
Codeman maintainer 5559aa8294 fix(remote): start csh and tcsh login shells without -i -l
A remote host whose login shell is tcsh came up on 1.48.0, when the local
sh rewrote ${SHELL:-/bin/sh} to the Codeman host's shell. Since #770 the
launch reaches the far end verbatim and runs tcsh itself, and tcsh refuses
-i -l -c, -l -c and -i -l, so every stock launch left a dead pane.

The login flags are now picked on the far end, in the shape of
loginShellArgs(): set -- -i -l, cleared for csh and tcsh, then
exec "${SHELL:-/bin/sh}" "$@". The text is one brace group, so
cd <path> && { ...; } still starts nothing when the cd fails. bsd-csh takes
every flag combination, csh is listed because it is tcsh on macOS and the BSDs.

The version probe keeps its one statement: the login shell parses it, not
/bin/sh, so fish and tcsh refuse it either way, as they did on 1.48.0.

Measured with the real builder on a real sshd in a throwaway Debian container
(tcsh 6.24.07, bsd-csh, fish 3.6, bash) and through a fake sshd with
tcsh 6.24.10, fish, zsh, dash and bash, a real Codeman server on tcsh included.
The tcsh disclosure comes out of the docs and the changeset. They keep the two
limits 1.48.0 had too: tcsh has no export, so the PATH bootstrap only prints an
error, and the tmux probe needs a command executable, which Debian lacks.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-06 06:05:53 +01:00
Codeman maintainer c65706a3e4 Merge #510 feat(adopt): adopt tmux sessions a human started, in all three locations
The attach line now takes the rule #770 gave remote and docker launches: bashCArg() single-quotes it at both respawn-pane sites, so it reaches bash verbatim. quoteAdoptLaunch is retired, since its double quotes kept the same $ and backtick safety but lost a real newline. #510's no-cd condition stays.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-06 05:36:20 +01:00
Codeman maintainer 44349395d0 Merge #753 feat(codex): let Codex sessions drive other sessions through the codeman skill
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-06 05:03:54 +01:00
Codeman maintainer aa40cca343 fix: adopted tmux targets, hung servers and the attach chain (release review)
Names tmux reads as target syntax (`@1`, `%1`, `=eq`, a bare `~`) passed the
has-session guard by name and then joined something else in new-session -t:
another session, or a fresh shell on the foreign server. They are refused at
discovery now, and take the existing skipped note.

One stopped tmux server emptied the whole local list and left a stuck client
per scan. Each server now gets a bounded ping with stdout at /dev/null first,
because a tmux client hands dups of its stdout to the server: a stopped server
holds our pipe open after the client is dead, so a watchdog around list-panes
cannot work. The probe runs in its own process group, all three transports
keep the rows read before a failure, and a missed ping becomes a note.

The attach and the probe pass -u, so a POSIX-locale container no longer garbles
the tab or lists a non-ASCII name as underscores. A docker attach gets the same
TERM and COLORTERM an owned container's exec gets.

A missing tmux no longer reads as "session is gone", and the launch line's `$`
now reaches bash unexpanded, so the ssh attach's PATH line is expanded at the
far end, as the probe's already was.

fix-plan/write, fix-plan/read and ralph-prompt/write refuse an adopted session:
its workingDir is the neutral /tmp, so they read and wrote a predictable name
in a folder anyone can write to.

FOREIGN_MAX_SOCKETS is gone: it capped nothing, and a real cap in glob order
would never reach `default` on a box with many `codeman-*` sockets (this one
has 212). Closing an adopted tab re-reads the outside-sessions list.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-06 04:33:42 +01:00
Codeman maintainer 353fe7ebcf fix: say what Allow network access really opens (release review)
The setting told users the sandbox and approvals stay on. With it on, a
command in Codex's sandbox can call Codeman's API, which can start a session
outside the sandbox with no approval prompt. Measured on codex 0.162.0 with
`codex sandbox`: a touch the sandbox refused landed through POST /api/sessions,
and from inside a Codeman pane it landed with a password set too, because a
pane inherits CODEMAN_PASSWORD. Without that password the same call gets 401.

The settings row, the wiki, the skill and the changeset now say so, the wiki
Security page and security-architecture name it, and the docs scope the claims
to what the code does: app-wide, local launches only, not cron jobs, and
--no-daemon is a local launch's flag. Agent-CLIs gains the --no-daemon
trade-off. A test pins that no surface sells the sandbox as still in charge.

Text only: the switch, the clamp and the launch are unchanged.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-06 03:25:00 +01:00
Codeman maintainer 131a12d195 docs(codex): say what Allow network access leaves alone
The owner's answer to the PR's open question: the switch does not also pass
--sandbox workspace-write. Settings, wiki, skill and release note now state
the consequence: a folder Codex runs read-only keeps the network closed and
asks per command. A golden test pins the decision by name, and the stale
launch-defaults comments now name networkAccess.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-06 02:52:52 +01:00
Codeman maintainer 57b2dcc466 Merge remote-tracking branch 'origin/master' into relprep/pr-510
Brings the 1.33.2 based adoption feature up to master. Conflicts were
resolved by keeping master's intent and re-applying the feature on top:
both sides kept for imports, the script tag and the stylesheet blocks; the
trust-dialog guard keeps master's stopped check plus the adopted refusal;
the Ralph and image watchers keep master's remote branch plus the adopted
skip; the close confirmation keeps master's worktree option plus the
adopted wording.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-06 02:51:00 +01:00
Codeman maintainer 35ff4e886c Merge origin/master into relprep/pr-753
Conflict in src/session-cli-registry-bridge.ts: kept the PR's gateVersion() (gates measured against the gated CLI's own version) and added master's Discord startup hint engine value.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-06 02:50:25 +01:00
Codeman maintainer b6eea702a8 fix: correct the pi wheel docs and comments (release review)
The gate comment in terminal-ui.js said a stale-false flag falls through
to the PageUp/PageDown fallback, which is false for pi since the second
commit of this PR removed that fallback. The FOOTGUN comment above it made
the same claim. Say what happens instead.

"Every pi below 1.0.0 emits no mouse enables" was too absolute: regular
was the default there, but fullscreen exists from 0.84.0. Measured with
--tui-mode fullscreen under tmux on 0.84.1, 0.85.0, 0.86.0 and 0.99.2:
alternate_on=1, mouse_any_flag=1, mouse_sgr_flag=1. Default settings on
0.99.2 report all 0.

Document that Wheel Scrolls Local History leaves the wheel dead in
fullscreen pi (wiki, integration guide, changeset), add a pi block to
Troubleshooting, cross-reference the accepted exception from the dead
gesture rule in architecture-invariants, and mark the pi-integration-plan
passages this PR reversed as superseded.

Comments and docs only, no runtime change.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-06 02:35:41 +01:00
Codeman maintainer b5aa3fbdf7 fix(pi): leave page keys out, and pin the strip that makes the wheel work
transcriptPageKeys applied to every pi, and a pi in `tuiMode: regular` binds
PageUp/PageDown to its editor. It emits no mouse enable, so the flag stays
false, and on a fresh pane the buffer is hollow: the wheel sent PageUp, the
caret of a multi-line draft moved, and the next keystroke landed on the wrong
line. Measured on real pi 1.1.0, 1.4.0 and 1.5.0. Regular mode is not exotic,
it only takes one line in pi's settings. Telling the two modes apart would
need a new capability, since the paging gate takes no id, so pi has no
fallback: a fullscreen pi with "Wheel scrolls local history" on stays where
master left it.

altScreen, the half of the fix that does the work, was pinned by nothing:
reverting it left 1328 tests green. The strip decision table reads its
expected value from the same registry entry. pi now has its own block using
the bytes it really emits, which also pins that focus reporting survives.

The fullscreen default dates from pi 1.0.0, not 1.4.0, so the comments say
that, and the sites still calling pi narrow-strip or main-screen by default
are corrected. Where a comment listed CLIs it now names the capability.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-06 01:04:01 +01:00
Randalix 58dae1e182 fix(pi): forward the wheel from cliMouseTracking so scrolling survives replay
pi 1.4.0 is an alt-screen TUI that enables mouse tracking once at start.
Codeman's replay (tab switch, reload) drops that mode in xterm
(mouseTrackingMode none, baseY 0), so the wheel hit an empty local buffer
and stayed dead until a resize made pi redraw.

Same shape as copilot: altScreen strip-mux-and-mouse (the server records
the stripped DECSETs as cliMouseTracking, which survives the replay),
wheelForward always, transcriptPageKeys (PageUp/PageDown fallback for a
stale flag, measured on pi 1.4.0). _shouldForwardWheelToApp treats pi like
copilot. Registry pins and docs updated.

Tests: 2 new in terminal-scroll-routing (red before). Full run: 12989
passed, 5 red from the de_DE locale / XDG env (custom-model-run-menu-ui,
git-worktrees, edit-history-sync), all green under LANG=C.UTF-8.

Co-Authored-By: Claude Sonnet 5.5 <noreply@anthropic.com>
(cherry picked from commit 53e4136d9f4401e9250ac6a122599ffcb2a57e8a)
2026-11-05 23:35:14 +01:00
Codeman maintainer 4960bee779 chore: one changeset for the release, and no new em-dashes
Consolidates the four contributor changesets, and rewords the em-dashes this release
added in comments and docs. The five in test/settings-save-failure-ui.test.ts stay:
they assert existing toast text.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-04 17:31:48 +01:00
Codeman maintainer 7c8d440928 Merge pull request #741: feat(discord): two-way Discord bridge with a thread per tab 2026-11-04 16:59:16 +01:00
Codeman maintainer 4f50af6a9f Merge pull request #740: feat(terminal): resolve relative path links against the pane's current directory 2026-11-04 16:57:30 +01:00
Codeman maintainer 8c317d84c8 fix: apply the review items for #755 at merge
Say that settings.yaml is shared only until the session's first settings write
(dsh saves by rename, which replaces the link), warn when a session's own DSH_HOME
does not answer and the server's home is linked instead, test
guardSessionConfigDirHome, and document the linkFromRealHome exception and what a
custom-model dsh session does not carry over.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-04 16:55:44 +01:00
Codeman maintainer 4c29aedbc5 Merge pull request #755: feat(custom-model): DeepSeek Harness injection through a per-session patch 2026-11-04 16:54:32 +01:00
Devvyn 82d251d52b docs(custom-model): align the docs and comments with the core behaviour 2026-11-02 16:54:57 +08:00
Devvyn fa2b81484b fix(custom-model): guard, document and test the session-supplied dsh home
Second review round on PR #755: skip non-absolute DSH_HOME candidates so a relative home can never mkdir under the server cwd; probe a session-supplied home with probePathKind in both async routes before the synchronous apply; test the session env-override lookup (seventh arg) incl. the self-link fall-through and the relative-home skip; reject '.' and '..' as realHomeFallback and enforce linkFromRealHomeDirs subset in the schema; fix the stale apply JSDoc and the two doc lines lagging the verified dsh run; state the shared/isolated split plainly.
2026-11-02 15:50:45 +08:00
Devvyn 1464332e91 fix(custom-model): create missing dsh home dirs before linking, rebuild restore bookkeeping, dedicated custom key var, schema guards for the real-home recipe
- linkFromRealHomeDirs marks directory entries: missing dirs are created (0700) instead of leaving a dangling link that failed every prompt with ENOENT; missing file targets are skipped
- _restoreCustomModel rebuilds the bookkeeping from the CURRENT recipe (configDir + env-key union) so a selection saved under the old env recipe can still be cleared after upgrade; drop the dead session-home recovery lookup
- deepseek's patch now references CODEMAN_CUSTOM_MODEL_API_KEY, so an in-session /model switch back to the official route cannot carry the custom key to api.deepseek.com (privilegedEnvKeys updated)
- schema: realHomeFallback must be one relative segment; linkFromRealHome requires it and must not collide with fileName
- docs: maintainer-verified real-dsh run replaces the outstanding-verification wording; shares-vs-isolates note; plan table row refreshed
2026-11-02 14:06:56 +08:00
Devvyn c646613c56 refactor(custom-model): move the real-home link into the registry recipe
- linkFromRealHome/realHomeFallback on the configDir recipe; the apply consumer is generic and names no CLI
- lstat guard so boot-recovery re-applies survive a dangling link; symlink is best-effort
- resolve the real home from the session's own dirEnvVar override before the server's, threaded through both routes and boot recovery (Session.envOverrideValue)
- deepseek links profiles/sessions/settings.yaml; boot-recovery test covers the session lookup
2026-11-01 21:42:02 +08:00
Devvyn aece02da87 feat(custom-model): DeepSeek configDir injection via a per-session patch
Replace the deepseek env redirect with a deepseek-patch-json recipe: a per-session DSH_HOME holding a generated cordis.patch.yml that declares a codeman-custom openai-completions provider over the installed profile. Ported from PR #752 commits 941df574 and 4bb1bfe9, minus the core context-limit changes.
2026-11-01 21:29:31 +08:00
Devvyn 1b8ac40baf fix(custom-model): core review fixes for context-length discovery and injection
- share one context-length resolution (discovered figure, endpoint default fallback) across the routes and boot recovery, with a recovery test
- record listed per-slot context figures with a slot count of 1 and force the slot-count fallback once per rediscovery
- open up the codex model launch token for qwen3-style ids (model-plain-colon), pinned by a spawn-golden case
- pass the full discovered window to opencode as limit.context
- translate the picker search strings and document the discovery precedence
2026-11-01 21:26:54 +08:00
Devvyn dbd1b6e587 refactor(custom-model): keep the DeepSeek patch rewrite on its own branch 2026-11-01 21:20:40 +08:00
Devvyn ed67596387 fix(custom-model): pass context limits and selected model to CLIs, and add model picker filter
- opencode: write limit.context/output from the discovered context length (#749)
- codex: override the native launch model when a custom endpoint is selected (#750), and set model_context_window
- model picker: add a search/filter box with a no-matches line
2026-11-01 21:06:38 +08:00
Codeman maintainer 87c656283d feat(codex): let Codex sessions drive other sessions through the codeman skill
Measured on 1.47.0 with the same "spawn a shell worker, echo, read, delete"
task: DeepSeek Harness and OpenCode already pass end to end; Codex 0.162
failed three ways, fixed here.

- Start Codex 0.162+ with --no-daemon (new noDaemonFlag gate, fail-closed).
  The shared app-server daemon stopped every launch on an "incompatible
  feature settings" dialog when another version's daemon was running, and
  ran commands with another session's CODEMAN_* environment.
- Evaluate registry version gates against the gated CLI's own version
  (resolveSessionCliVersion) instead of claude's for every entry.
- Add the opt-in synced setting App Settings > Codex > Allow network access
  (codexNetworkAccess, default off): --config
  sandbox_workspace_write.network_access=true, filled as a launch default
  and clamped off for non-granted owners like the bypass switch.
- Skill: carry on with the curl path when `codeman` is not on PATH, re-run a
  sandbox-blocked call once through the CLI's approval path, name the new
  setting. Wiki and registry docs updated.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-11-01 06:21:38 +01:00
Devvyn 1c4338e8b5 feat(discord): codeman notify, startup hint for every CLI, and file send/receive
Adds the codeman notify command, a startup hint (Claude via --append-system-prompt, other CLIs once on first idle), files saved from thread attachments into the tab folder, and --file uploads limited to the session folder.
2026-10-29 21:27:23 +08:00
Aamer Akhter f23e18b1fc fix(terminal): report the pane directory in the workspace's own spelling
tmux reads a pane's directory from the process cwd, which is always the
resolved path, while workingDir keeps the spelling the case was created
with. For a workspace reached through a symlink every reading looked
like it was outside the workspace, so the browser sent each relative
link through the attachment route instead of the workspace routes.

Each session now resolves workingDir once in the background, bounded by
probePathKind() and a timed realpath(), and setCwd() writes a reading at
or under the resolved path back in workingDir's spelling. Until that
settles only a reading lexically inside workingDir is taken. An unknown
probe or a failed realpath leaves readings untranslated and retries
after WORKSPACE_REALPATH_RETRY_MS.

Also corrects two comments: cwd is written to state.json but never read
back, and a blank pane path is usually a pane whose program exited.
2026-10-28 19:29:05 -04:00
Codeman maintainer 98526527c8 Merge pull request #711: fix(mobile): hold the PTY rows while the on-screen keyboard is up 2026-10-27 02:24:17 +01:00
Codeman maintainer 5908cc51fd Merge pull request #694: feat(agent): COD-529 add bounded inspection and guarded submission
Conflicts resolved by keeping both sides:
- CLAUDE.md (Agent CLI): #694's opening (an HTTP client, the experimental
  agent-session-routes behind agentCoordinationEnabled) with #708's clause
  on the directory spawn and its two refusals.
- docs/api-reference.md (codeman agent): #708's spawn paragraph and --json
  shape, with #694's inspect/stage/submit in the verb list and the guarded
  endpoints named before "adds no route for the other verbs".
- src/cli-agent.ts (@fileoverview): #708's quick-start / directory spawn
  routes and #694's experimental routes in one endpoint list; the two
  node:fs/promises imports (#708's realpath+stat, #694's open+stat) merged
  into one.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-27 02:14:44 +01:00
Codeman maintainer 6539bc6c82 fix: apply the review items for #714 at merge
- Session update no longer resets "How often" while typing: updateGitFetchStatus
  (session-ui.js) leaves a control alone while it is document.activeElement, and
  autoSaveGitFetch writes the clamped interval back itself (Enter saves with the
  field still focused).
- Partial override shows the app values: enabled = session ?? app switch,
  interval = session ?? app interval ?? 300 (session-ui.js updateGitFetchStatus).
- runFetch (git-fetch.ts) arms the spawn with what the core.sshCommand read left
  of timeoutMs, and answers ETIMEDOUT without spawning when nothing is left, so a
  git that never answers costs one timeout, not two.
- fetchEnvFor keeps a GIT_SSH from the server's environment (repoSshCommand ||
  base.GIT_SSH), fileoverview and CLAUDE.md line updated to say so.
- i18n.js: the zh-CN "Git fetch on an interval" description scopes the only
  writing git command to the Git window.
- styles.css: the disabled/busy rules are scoped to .git-status-actions
  .btn-icon-sm (moved beside .git-status-actions), and the busy pulse stops
  under prefers-reduced-motion.
- Comments that called the override per case now say per session: session.ts
  (field, restore and getter docs, setGitFetchSettings), git-fetch-scheduler.ts:2,
  server-timing.ts:78, server.ts (cleanup and restore), app.js _onSessionUpdated,
  session-ui.js (openSessionOptions and updateGitFetchStatus).
- Tests: test/git-fetch-session-options-ui.test.ts (new, JSDOM: typing survives
  an SSE update, the Enter write-back, both partial cases); test/git-fetch.test.ts
  (GIT_SSH kept; one bound for the read and the fetch, stub git on PATH).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
2026-10-27 01:58:14 +01:00
Aamer Akhter 2832a037d6 fix(agent): verify staged text on a live codex pane and refuse popups
Stage now re-reads the pane after typing (every 50 ms, up to 1 s) and
issues a receipt only for a frame with the same generation, the input
revision its own write moved, and a composer hash equal to the
message's. Codex repaints 30 to 60 ms after the keys, so the first
capture was often stale and stage refused with the text already typed.
Every refusal after typing now says the message may be in the composer
and points at the Ctrl+U recovery. A visible-frame capture reads the
screen and the caret in one tmux invocation, with the old two calls as
the fallback.

Codex draws its suggestion popups above the composer and keeps the
footer, so a staged /stat verified and submit ran /status. Stage now
refuses a leading / or ! and a last word starting with @ or $ before
typing, and stage and submit refuse a frame with popup rows just above
the composer. The pattern and characters are registry data
(composerPopupLine, composerCommandPrefixes, composerCompletionSigils),
read from the codex 0.162.0 bottom_pane sources.

Submit waits until 250 ms after the staged keys: codex turns an Enter
within 120 ms of a typed burst into a newline. POST .../input answers
409 with its seq returned while a guarded submit holds the input lease,
instead of 200 for bytes it never wrote.

Also align the attribution memo comment with the code (a verified file
whose incremental scan fails is not re-walked, since the full walk
reads a superset), and list stage as a fromUser source in CLAUDE.md.
2026-10-25 09:22:35 -04:00
Devvyn 9432d9f9e2 feat(git): a Fetch button in the Git status header, and an interval for it app-wide and per case
Fetch is the only git command Codeman runs that writes, and the only one that runs a
repository's own fetch-post hook, so it stays behind a click and off by default. The Git
status window's header row gains Fetch; App Settings -> Bottom bar gains the switch and
the interval, synced because the scheduler acts on them, and Session Options -> Git
overrides both per case. The scheduler reads settings fresh each tick, times the interval
from the start of a run, never overlaps runs, and re-checks the Docker and remote rules
every tick before any git runs. Nothing but remote-tracking refs moves: HEAD, the index
and the working tree are proven unchanged against real git while origin/main advances.

Tests: src/git-fetch.ts and the scheduler, the per-case settings route, the synced/display
setting split, the two new i18n rows, and the Fetch block of the Git status browser suite.
2026-10-25 17:23:55 +08:00
Aamer Akhter c78635a3d3 feat(agent): make coordination routes opt-in and address review
Gate inspect, stage and submit behind the synced agentCoordinationEnabled
setting (default OFF, App Settings > Cross-session features, zh-CN
strings). While off, all three routes answer 403 FORBIDDEN naming the
setting, checked before session lookup; the CLI verbs print it verbatim.

Remember the attributed Codex rollout per session and re-verify only its
head on later calls, rescanning only date directories not older than it
for a /new conflict. The first walk reads 1 KiB prefixes first, so a
store of 2,100+ unrelated rollouts still attributes.

Answer unexpected inspection errors with 422 OPERATION_FAILED and log
the session id and error code only.

Read the guarded composer glyph and footer from the CLI registry: codex
workDetect.promptGlyph and a new workDetect.composerFooterLine
capability, compiled through compileVersionRegex. Codex footer effort
words are one shared list with modelDetect.screenLine.

Fix the wiki API reference link, frame the versioning policy entry as a
case of the experimental rule, join the CLAUDE.md Agent CLI paragraph,
and update the changeset, skill reference and architecture notes.
2026-10-24 22:54:17 -04:00
JD 3e334357e8 fix(mobile): hold the PTY rows while the on-screen keyboard is up
Every keyboard open and close resized the PTY, and Claude Code's inline renderer answers each height change by clearing the screen and reprinting only the transcript tail that fits, so rows above that tail were lost from tmux for good. Measured on Claude Code 2.1.306 with tmux 3.2a: five keyboard cycles (62x38 to 62x15 and back) left 75 of 100 numbered lines in the pane's history; with the rows held the PTY stays 62x38, no SIGWINCH reaches the pane and all 100 remain (#672).

While the keyboard is up the rows captured at the show transition are held through getTerminalDimensions, columns still follow the box, and the screen, now taller than its box, is translated up by --kb-clip so the last content row sits at the bottom of the visible area. This is the one documented exception to the columns-only geometry rule: there the extra rows came from another device and nothing kept the composer in view, here they are this device's own rows held on purpose and the clip keeps the composer visible. The local-echo overlay's setRowLimit narrows the renderer's existing totalRows bound, so one place decides where a long draft stops growing downward.

Only inline renderers hold, through a new registry capability keyboardHold: 'inline' (Claude alone) published to the page as window.__codemanKeyboardHoldModes. Full-height layouts such as codex, Claude's fullscreen mode (mouse tracking on, unmeasured), and remote or docker panes resize as before. The hold is released on keyboard close, a shape change, a tab switch, a session update, a reconnect, or the setting turned off. Per-device setting Keep Size Under Keyboard, default on.
2026-10-24 20:17:45 -04:00
Codeman maintainer b7897b6d0e fix: apply the review items for #684 at merge
Multi-user: add a (username, IP) failure bucket checked before verifyPassword, capped at AUTH_FAILURE_MAX, with the same decay as the other buckets; the per-IP pre-gate stays at AUTH_IP_FAILURE_MAX and the per-username bucket stays after the verify
A correct password now also clears that account's (username, IP) bucket, beside the IP and per-username buckets
The new bucket is disposed on server stop like the others; the comments above the gate and the constants' JSDoc describe the three buckets
Tests: 10 wrong then the correct password from one IP is a 429, the same account logs in from another IP, another account behind that IP is unaffected, and a correct password clears the pair bucket
test/multiuser-auth.test.ts: the spray test loops to AUTH_IP_FAILURE_MAX instead of a literal 50
Reworded the rate-limit lines in auth-config.ts, security-architecture.md, README.md, README.zh-CN.md, wiki/Security.md, endpoints.md (plus the plugin mirror), CLAUDE.md and architecture-invariants.md
docs/wiki/Remote-Access.md: add the multi-user note to the tunnel page

Co-Authored-By: Claude Opus 5.5 (1M context) <noreply@anthropic.com>
2026-10-24 14:15:21 +02:00
Aamer Akhter 175e3b8320 fix(auth): keep the single-user login throttle at 10, widen only the multi-user pre-gate
Single-user mode has no per-username bucket, so its per-IP bucket is the
only limit on guesses against the one password, and behind a tunnel every
client shares 127.0.0.1. Put it back on AUTH_FAILURE_MAX (10).

AUTH_IP_FAILURE_MAX (50) now gates only the multi-user per-IP pre-gate,
which runs before the scrypt verify and so refuses even a correct password;
the per-username bucket still caps each account at 10.

Move the headroom test out of the single-user suite and add a multi-user
test that sprays wrong passwords across usernames (9 each) and expects 401
for the first 50, then 429 for the 51st and for a correct password.

Update the docs that still described 10 per IP for every mode, and resync
the plugin mirror of the codeman skill.
2026-10-22 21:21:49 -04:00