agentleFS
Seth090502Seth090502/osanwe-public/.agents/skills/gate/SKILL.md

gate

Skill0 starsChanged 53 days agoMITView on GitHub

Run a judgment gate before acting: /gate f trade discipline (anti-FOMO / anti-panic) before any ADD / TRIM / EXIT, /gate b build-vs-ship before building a new skill, script, hook, workflow or index, /gate t before changing a thesis status on a manual trigger (did it FIRE?), /gate calibrate for the routing-around compliance report. The model never judges -- verdicts come ONLY from tools/gate-eval.py. Not for sizing or doctrine-band questions.

33 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use new, build and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • codebycoreySkill

    new

    Start a new change with a proposal. Use when starting a feature, creating a proposal, or beginning planned work.

    Also in 3 projects that use this

  • Microsoft AzureSkill

    build

    Build the TypeSpec Rust emitter from source. Run this after cloning the repo, after pulling changes, or when TypeScript source files have been modified.

    Also in 5 projects that use this

  • AWSSkill

    ship

    Finalize dev work on an issue branch — update docs, push, and open a PR.

    Also in 8 projects that use this

  • genomewalkerSkill

    checkpoint

    Save a checkpoint of task status, decisions, active files, and next steps

    Also in 3 projects that use this

  • VercelSkill

    investigate

    Investigate an issue report, or analyze the constraints of a proposed change, with strict, evidence-driven diagnosis.

    Also in 3 projects that use this

  • BlockSkill

    plan

    Draft a TDD, PRD, or lightweight plan in the conversation without committing it.

    Also in 4 projects that use this

The file

We split this skill into the rules it gives an agent

4 rules. 0 are common, 3 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 30 edited versions we read keep these.

  • 3

    Distinctive

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Every marker = {value, prov} per the fill-rules in ref-gate-tables.md.

    line 43
  2. Distinctive:

    NEVER estimate a marker you can read.

    line 45
  3. Distinctive:

    If the verdict surprises you, the markers are wrong or the table is wrong -- say so to the user.

    line 51
  4. Distinctive:

    Every marker carries provenance.

    line 97

“Common” is measured: kept by at least 6 in 10 of the 30 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

/gate <f|t|b|calibrate> <subject> [--preview]

One artifact set encodes the judgment; this skill only fills forms. AUTHORITATIVE rules: tools/gate-rules.json. Human mirror + fill-rules + worked examples: ref-gate-tables.md (sibling of this file) -- READ IT BEFORE FILLING ANY MARKER. Golden exemplars (imitate them): the golden: true sheets in wiki/research/gates/.

Mode routing (Pattern 6 -- deterministic, from args only)

  • f <ticker + action> -> GATE-F trade-generation discipline sheet
  • t <thesis> <trigger-id> -> GATE-T trigger-fired adjudication sheet
  • b <proposal> -> GATE-B build-vs-ship sheet
  • calibrate -> compliance report only (no sheet)
  • --preview on any mode: run Phases 0-B, print the would-be sheet, write nothing.

Phase 0: Load context (read-only)

  1. tools/gate-rules.json + ref-gate-tables.md.
  2. wiki/research/gates/gates-registry.md (prior sheets on this subject; a BLOCKED verdict on the same instrument within 5 trading days forbids a re-gate -- HALT).
  3. Mode inputs: F -> entity note wiki/entities/tickers/<T>.md, latest analysis in wiki/investing/analyses/, latest briefing meta.json in Calendar/decisions/briefings/ (VIX fields), Calendar/decisions/execute-or-decline.md; T -> the thesis file's triggers: block + the evidence sources themselves; B -> execute-or-decline.md + open-loops digest + the nearest similar artifact.

Phase A: Fill markers

Every marker = {value, prov} per the fill-rules in ref-gate-tables.md. Provenance: vault-relative path, or a scheme prov (mcp:*, script:*, web:*, manual:*). NEVER estimate a marker you can read; NEVER omit one. Sheets carry NO dollar amounts, share counts, or position sizes -- fractions and percentages only.

Phase B: Compute (never judge)

python tools/gate-eval.py --compute <draft-sheet> --json Transcribe verdict, mandates, suggested review_date VERBATIM into the sheet. If the verdict surprises you, the markers are wrong or the table is wrong -- say so to the user; do not adjust markers to steer the verdict (that is the failure mode this kit exists to stop).

Phase C: Write the sheet

wiki/research/gates/gate-<mode>-<slug>-<YYYY-MM-DD>[-HHMM].md (archival rule; never overwrite). Frontmatter: canonical + gate: block exactly as in the goldens; related: includes [[gates-registry]] + the entity/thesis touched. Body: two sections -- ## Evidence (prose behind each marker, cited) and ## Verdict + mandates (restated; user overrides are recorded HERE with the override reason -- markers are never edited to flip a verdict).

Phase D: Check (HALT gate)

python tools/gate-eval.py --check <sheet> --json Exit 2 -> HALT: report findings, fix the sheet, re-check. Never proceed unchecked.

Phase E: Registry row

Append one row to the index table in wiki/research/gates/gates-registry.md: | date | mode | subject | verdict | mandates | review_date | outcome | (outcome starts empty; a later session fills VINDICATED / WRONG / OVERRIDDEN).

Phase F: Execute mandates

  • eod-row-* mandates -> append the row to Calendar/decisions/execute-or-decline.md per its row protocol (ID = next EOD-N, ESCALATION_DATE from the mandate).
  • status-floor-* -> the existing /brief Pattern-20 path does the fired: stamp; this sheet is its evidence record -- reference the sheet in the briefing, do not edit thesis files from here.
  • Report every mandate executed / deferred in the final output.

Phase G: Daily note

One line to today's Calendar/daily/ note under the session's actions: - [HH:MM] /gate <mode> <subject> -> <verdict> plus a Cross-References wikilink to the sheet.

Mode: calibrate

python tools/gate-eval.py --calibrate --since <date> --json (default 30d). Append a dated ## Compliance <date> block to gates-registry.md with the per-gate expected/gated/compliance table, misses, ROUTING-AROUND flags, and mandate follow-through gaps. Surface flags to the user bluntly -- low compliance means the gates are being routed around, which is the primary threat model. Also fill any outcome: fields whose review_date has passed (judge from what the vault records show happened; OVERRIDDEN when the user acted against a verdict).

Quality Rules

  • Verdicts come from gate-eval.py output only; transcription must be verbatim.
  • Every marker carries provenance; vault-path provs must exist on disk.
  • ASCII only (Pattern 22). No dollar values / share counts on any sheet.
  • Registry row per sheet, no exceptions -- unindexed sheets defeat calibration.
  • Empty-input safeguard: if the subject cannot be resolved to a concrete instrument / trigger-id / proposal, HALT and ask -- do not gate a vague subject.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.