The implement-plan flow asked to commit the closed plan together with
the code, so the resulting commit looked like it was missing a piece.
State the rule in the three places that describe it: the implement-plan
skill, the commit workflow memory and the agents README. A plan is
session bookkeeping, not part of the change, and its absence from the
commit is correct rather than a missing piece.
AI-assisted-by: deepseek-v4.1-flash
The new skill turns HEAD's stacked subjects, bodies and repeated trailers
(a typical result of squashing) into one message that follows
mem:workflow/creating-commits: a single imperative subject, a body
covering the whole branch diff, then the trailers.
It collects every distinct AI-assisted-by value before rewriting, so an
attribution is never dropped, and it requires ./scripts/check-commit to
exit 0 after the amend. It never pushes: rewriting history over commits
the user already pushed is their call, not the agent's.
AI-assisted-by: deepseek-v4.1-flash
Relax the subject limit from 70 to 80 characters and the body
wrap limit from 76 to 80. Merge/Revert/Reapply commits skip the
subject length check since git prepends those prefixes to an
existing subject. Update the checker, its tests, and all related
docs and memories accordingly.
AI-assisted-by: muse-spark-1.3-contributor
Signed-off-by: Andrey Antukh <niwi@niwi.nz>
The Co-authored-by trailer is now forbidden. The
commit checker fails any message holding it, whatever
the case, and the docs record the ban.
AI-assisted-by: muse-spark-1.3-contributor
Signed-off-by: Andrey Antukh <niwi@niwi.nz>
The review-code flow skill now documents how to post a review as a
GitHub comment: in English, with a minimal structure (intro line,
main issues, minor suggestions, optional closing). Posting is
explicit-only: the review stays in the session unless the user
asks for the comment.
The code-review-criteria skill replaces the flat finding ID
sequence (F1, F2, ...) with severity classes: R1 for required
findings, M1 for minor ones, and S1 for suggestions. IDs stay
stable so follow-up discussion can reference findings
unambiguously.
AI-assisted-by: glm-5.3-flash
* ✨ Take changelog highlights from the release highlight label
The 🚀 section was picked by the agent on each run and kept
forever, so a regeneration could add highlights product did not choose
and nobody could drop one. Make the `release highlight` label the only
source, as `breaking change` already is for 💥, and have the report
flag 🚀 entries that do not match the label.
AI-assisted-by: claude-opus-5-5
* 📚 Update changelog for 2.19.0
Regenerate the 2.19.0 section from the milestone with the
update-changelog skill. Add the 79 closed issues that were missing,
drop entries labelled `no changelog` and #9263, which is not in the
milestone, and take 🚀 from the `release highlight` label.
AI-assisted-by: claude-opus-5-5
Check a milestone against the Main project board with
scripts/project-anomalies.py (check writes
tmp/<MILESTONE>-ANOMALIES.md): open issues with merged PRs,
milestone mismatches either way, needs triage labels and
unassigned non-community issues. Outsiders resolve in bulk;
only a missing milestone auto-fixes, on confirmation.
Teach scripts/gh.py batched issue lookup (issue subcommand),
milestone on prs output, plus assignees and projects fields,
and retry transient HTTP 504s.
AI-assisted-by: muse-spark-1.3
Move the inline programs out of the update-changelog skill into
scripts/changelog.py (check-merged, cross-ref, report) and shrink
the skill to the workflow alone.
Teach scripts/gh.py to retry transient HTTP 504s, resolve lookups
in batches of 50 (new issue subcommand, milestone on prs output),
and report 💥 entries without the breaking change label.
Add the nine missing 2.17.0 changelog entries found while
trying the new flow.
AI-assisted-by: muse-spark-1.3
* 🎉 Move backend jobs to the unified job table substrate
Replace the generic 'task' table with a unified 'job' table as the
durable substrate for all backend jobs, and migrate every producer
and consumer to it. The legacy 'task' table stops receiving writes
but stays dormant (still cleaned by tasks-gc); upload-session
chunking and the other post-branch upstream changes are preserved.
New 0154 migration creates the 'job' table: dispatch and lifecycle
columns (name, queue, priority, scheduled_at, retries, status,
timestamps, props) plus optional user-facing ledger columns
(profile_id, target, progress, error, result, resource_id,
expires_at). No modified_at trigger: the application updates it on
every write for lease and orphan detection. Partial indexes cover
the dispatcher claim query, orphan scanning, and per-profile
listing.
New app.jobs public API: submit!/invoke!, heartbeats and progress
reporting, a precompiled job-def registry (name, schema, handler,
decoder, validator), an ephemeral request/response mode, and a
management API for external workers. The dispatcher claims
new/retry rows (FOR UPDATE SKIP LOCKED) with a plain JSON redis
payload and marks lease-expired running rows as orphaned; the
runner executes per-queue with the same retry conventions.
New jobs GC (daily): deletes expired rows and retains internal
terminal rows per :jobs-retention, touching referenced storage
objects so storage-gc-touched reclaims them. Job resources live in
a dedicated job-resource bucket with a reclaim branch in
storage-gc-touched.
All production call-sites (sendmail, delete-object, demo-purge,
file-gc, offload, webhooks, quotes, nitrate bulk delete,
snapshots, telemetry) and the ten cron tasks now submit through
app.jobs; cron is a pure scheduler and the legacy worker registry,
task-table inserts, and per-namespace handlers are removed.
Includes migrations, dispatcher, runner, jobs, request/response,
jobs GC, management API, and cron test suites.
AI-assisted-by: glm-5.3-flash
AI-assisted-by: muse-spark-1.3-contributor
AI-assisted-by: kimi-k3
AI-assisted-by: mimo-v2.5
* ✨ Address review comments on jobs substrate
Resolves the four open review threads on the jobs PR. Orphaned jobs
are now marked aborted, a system-side terminal state with no retry:
the dispatcher sweep stores the end event with outcome aborted in a
single bulk insert in the same transaction, records the terminal
outcome next to the orphaned counter, and reports an error log;
orphans keep alerting through the log and the counter, with no
user notification.
The aborted state is terminal everywhere: the status CHECK, the
sweep, the lifecycle writers, the jobs GC retention, the backlog
gauges and the user status mapping, with tests covering each
transition. The claim comment, the 2.20 doc version and the legacy
task table notes are corrected, and objects-gc heartbeats per chunk.
AI-assisted-by: muse-spark-1.3-contributor
Expose selected agent skills as slash commands and pass invocation
arguments to their entry points.
Update the devenv OpenCode CLI to v2.0.16 with checksums for both
supported architectures.
AI-assisted-by: space-bunny-free
Clarify that plan files stay local and that only implementation,
tests, memories, and documentation belong in the final commit.
AI-assisted-by: Space Bunny Free
GitHub does not report mutation-created issue-to-PR links through
closedByPullRequestsReferences(userLinkedOnly: true), so the
verification in `gh.py link-issue` failed even when
addCloseIssueReferences succeeded and the link existed.
Drop the re-query and trust the successful mutation: the command now
fails only when a link target is missing or the mutation does not
return the issue. Update the tests, the gh helper memory, the PR/issue
workflow memories, and the create-pr skill so they no longer promise
verification.
AI-assisted-by: deepseek-v4.1-flash
Add a link-issue command that creates GitHub's Development reference
and verifies both sides. Keep Closes in descriptions for context, but
make the API link the source of truth, including for merged PRs.
Add tests for successful links, missing verification, output, and
failures. Update the PR workflow memories and create-pr skill to use
the command.
AI-assisted-by: space-bunny-free
* ♻️ Keep agent content on standard paths, drop client shims
Agent guidance stays where the ecosystem reads it: AGENTS.md at the
root and the skills in .agents/skills. Codex scans .agents/skills and
follows symlinks, opencode lists it among its discovery locations, and
Cursor, Zed, Warp, Cline and Amp use the same project path.
Claude Code reads neither, so it needed CLAUDE.md and .claude/skills.
Committing those two paths is not a neutral default. Both are
gitignored, git treats ignored files as expendable, and a checkout
that needs the path deletes whatever a contributor keeps there without
a warning. Neither sparse-checkout nor skip-worktree offers a way out,
because both make the next pull abort. The two links are now created
by the contributor, once per clone, and documented in
.agents/README.md.
AGENTS.local.md and .agents/local/ give every client one personal
layer that composes on top of the shared one, instead of one file per
client inside the repository.
sync-workspace seeds the two links for ws1 and above, the way it
already seeds the gitignored config.js, so the devenv workflow does
not change.
AI-assisted-by: claude-opus-5
Signed-off-by: Álvaro Tejero Cantero <alvorithm@teje.ro>
* ♻️ Lean on Claude Code reading AGENTS.md, drop the pointer
Claude Code 2.1.277 reads AGENTS.md through the built-in agents-md
plugin, whose default mode loads it in a project that has no
instruction file of its own. Deleting the committed CLAUDE.md is what
turns that on here, so the pointer a Claude user had to create is no
longer part of the setup, and only the .claude/skills link remains:
the plugin covers instruction files, and skills still load from that
path alone.
The instruction that replaces the pointer is a removal. A CLAUDE.md,
.claude/CLAUDE.md or CLAUDE.local.md anywhere from the root down to
the working directory hands the project back to CLAUDE.md and hides
this repository's AGENTS.md, while .claude/rules/*.md and the
person's own ~/.claude/CLAUDE.md do not count and stay usable for
personal steering. CLAUDE.local.md joins the ignore list, and a hard
rule in AGENTS.md forbids all three.
The docs name the version, say the fallback has not reached Bedrock,
Vertex or Foundry, and say the stable auto-update channel runs about
a week behind the default one, so the symlink is documented as the
stopgap for those cases and as something to drop afterwards.
sync-workspace seeds the skills link alone for the same reason.
AI-assisted-by: claude-opus-5
Signed-off-by: Álvaro Tejero Cantero <alvorithm@teje.ro>
* 📎 Remove skills directory mention on AGENTS.md
Signed-off-by: Andrey Antukh <niwi@niwi.nz>
---------
Signed-off-by: Álvaro Tejero Cantero <alvorithm@teje.ro>
Co-authored-by: Andrey Antukh <niwi@niwi.nz>
Give every plan a write-restricted Status (draft, reviewed,
done) and an append-only Review Log with UTC ISO 8601 lines.
Make-a-plan creates plans as draft and is the only flow
writing reviewed, on explicit user apply. Review-plan stays
read-only. Implement-plan closes the plan to done with the
issue URL when one exists, in the same commit as the code.
Document the lifecycle in the agents README.
AI-assisted-by: muse-spark-1.3-contributor
Define derived plan naming for .agents/plans/.
Parent basename stays intact and derivatives append
--review-NN for review followups and --task-NN for
roadmap sub-plans, with no new date so ls groups them.
Document the rule in the planner skill, the in-place
vs new-file policy in make-a-plan, and examples in
the agents README.
AI-assisted-by: muse-spark-1.3-contributor
Add the verified REST procedure for linking an issue as a
sub-issue of an umbrella/EPIC: get the REST id, POST to the
parent's sub_issues endpoint with a typed -F field, and verify
both directions. Route it from the create-issue skill.
AI-assisted-by: deepseek-v4.1-flash
Delegating plan and review to a subagent (engineer-* or the
builtin general) starts a clean context instead of growing the
main session. Delegating to general keeps the same model.
AI-assisted-by: muse-spark-1.3-contributor
Turn the skills-only file into a full intro to opencode inside
plain devenv: setup, providers, models, opencode.json example,
gh auth, flows, and a skills summary at the end.
Provider, model, and flow sections follow Andrey's own setup
notes; the FAQ stays as a stub for later.
AI-assisted-by: muse-spark-1.3-contributor
Add a body line-length validator to scripts/check-commit. It
fails when a body line exceeds 76 characters, exempting
trailers, URLs, and unbreakable tokens. The 76 limit leaves
room for git log's four-space indent in an 80-column
terminal.
Align the subject limit with the documented 70 characters;
the checker allowed 90 before.
Document the rule as a hard, verifiable requirement in
AGENTS.md, CONTRIBUTING.md, the create-commit skill, and
the workflow memory, and point at scripts/check-commit.
Add tests for the validator and the subject length rule.
AI-assisted-by: deepseek-flash
Teach agents to verify their changes with ./scripts/ci: module list,
task selection flags, log locations under .ci-logs/, and per-module
workflows (lint-only pass, --fix, --paren-repair, common/ consumer
checks). Register the skill in the skills README, point AGENTS.md at
the new memory, and add the script to the critical-info dev scripts.
AI-assisted-by: omen-alpha
The flow only supported continue mode (existing branch) or standalone
mode (issue + branch). Add a "direct" argument (aliases: "no branch",
"direct commit") that skips both and commits on the current branch as
it is, even when it is a base branch — for small tooling changes the
user wants committed in place.
AI-assisted-by: omen-alpha
Two changes to the implement-plan flow:
- It now shows the whole picture before touching the repo: which
issue and branch will be created (or the current branch it
continues on), the execution style, and the task checklist — then
waits for the user's confirmation. Flow detection is read-only;
issue and branch creation moved to the execution step.
- New opt-in execution mode: on request ("step by step"), it runs
one task at a time, commits each one with the create-commit skill,
and waits for the user's review before the next task. The default
stays as it was: every task, one commit.
README and the implement-plan command description updated to match.
AI-assisted-by: omen-alpha
.agents is the shared home that opencode, Claude Code (through the
.claude/skills symlink) and Codex all read, so the skills and the
saved plans now live there instead of .opencode:
- .opencode/skills moved to .agents/skills (24 files, no content
changes).
- .opencode/plans moved to .agents/plans; the .gitignore entry
follows, so plans stay untracked.
- .claude/skills symlink retargeted to ../.agents/skills.
- planner, make-a-plan and review-plan updated to the new plans path;
new .agents/README.md documents every skill with when-to-use
examples and links to each SKILL.md.
- workflow/creating-issues memory: create-issue path updated.
opencode discovers .agents/skills natively, so .opencode needs no
reciprocal link.
AI-assisted-by: omen-alpha
