bulma-lstm
Skill1 starsChanged 50 days agoMITView on GitHub
Graph: author-side of review. Ingest a PR, review, inline thread, or issue-comment URL, verify with receiving-code-review, patch should-fix on the same branch, and reply on every comment. Use when /lstm, /bulma-lstm, CHANGES_REQUESTED, or LSTM_REQUEST.
In this repository35 other files it gives its agents.
- AGENTS.mdAGENTS.md
- skills/bulma-feature-delivery/adapters/claude.mdCLAUDE.md
- architectSkill
- before-and-afterSkill
- blast-radiusSkill
- bulma-busSkill
- bulma-code-reviewSkill
- bulma-developerSkill
- bulma-fdSkill
- bulma-feature-deliverySkill
- bulma-goalSkill
- bulma-hostSkill
- bulma-memorySkill
- bulma-qaSkill
- bulma-reviewerSkill
- bulmaSkill
- bulma-triageSkill
- diagnoseSkill
- figure-it-outSkill
- grill-with-docsSkill
- howSkill
- interrogateSkill
- kevinSkill
- no-commentsSkill
- principle-fix-root-causesSkill
- principle-prove-it-worksSkill
- principle-sequence-verifiable-unitsSkill
- receiving-code-reviewSkill
- tddSkill
- technical-writingSkill
- thermo-nuclear-code-quality-reviewSkill
- to-issuesSkill
- to-prdSkill
- typescript-best-practicesSkill
- unslopSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- could be automatic: a setting could enforce it
- Distinctive: line 16
Never merge. Same PR. Same branch.
- Very specific: line 38
ASK last resort: DECISION_POLICY.md.
- Very specific: line 54
Do not spawn bulma_developer / bulma_reviewer / bulma_qa.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Bulma LSTM (graph)
Looks shit to me. Orchestrator for incoming review, not /bulma-reviewer.
Never merge. Same PR. Same branch. Draft stays draft. No new PR.
REQUIRED: GRAPH.md · ARGS.md ·
STATE.schema.md ·
bulma-host ·
receiving-code-review ·
kevin · bulma-memory ·
GITHUB.md ·
bus PROTOCOL.md ·
DISK.md ·
fd DECISION_POLICY.md
Init .bulma-lstm/STATE.md. Walk GRAPH (admit first).
Busy main or 2+ PRs → worktree + general-purpose worker per PR.
Worktree and branch rules: JOBS.md §Worktree.
Orchestrator does not write product code. patch spawns
bulma-fd-coder per evidenced cause group (TDD), preserving per-comment acceptance. prove runs the fd reviewer (spec + quality)
and tester before reply. A human sign-off request also requires a bus
switch to bulma-qa for visual behavioral evidence, even without a code fix
(request_qa.md). QA evidence supports the human decision;
it never grants human approval. Grill only when the fix is complicated.
ASK last resort: DECISION_POLICY.md.
Chat: bulma-memory. Apply Kevin for human-facing communication. Every GitHub reply
(thread, skip reason, COMMENT review, re-review request) is English,
rewritten with bundled kevin before POST. Never POST the first draft.
Reply is not optional. Every analyzed comment gets 👍 and a thread reply. After proven code changes, post a summary, resolve fixed conversations, dismiss fully addressed CHANGES_REQUESTED reviews with evidence, and re-request original reviewers unless their latest decisive status or the PR is APPROVED. Skip-only runs reply and summarize without dismissing or re-requesting. See reply.md. Then learn.md records verified causes, scoped prevention for coder/reviewer/QA and eligible improvement task drafts. LSTM invocation alone does not prove a miss. Issue publication follows FEEDBACK.md, defaulting to local drafts.
Do not spawn bulma_developer / bulma_reviewer / bulma_qa.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.