build-log
Skill0 starsChanged 4 months agoMITView on GitHub
Use when a phase completes or a sanity-gate result is being fixed as permanent evidence — writing a new numbered entry under the build-log directory. The bar is high; single-module additions and small changes do not get entries. Trigger phrases: "phase N is done, log it", "the gate passed, record it", "write a build log entry", "completion note", "fix this result as evidence".
In this repository15 other files it gives its agents.
- templates/CLAUDE.mdCLAUDE.md
- canonical-paramsSkill
- correctness-gateSkill
- decision-logSkill
- evidence-conventionSkill
- layer-syncSkill
- phase-gateSkill
- project-layoutSkill
- run-provenanceSkill
- external-positioningSkill
- lit-anchorSkill
- negative-results-ledgerSkill
- numerical-determinismSkill
- statistical-reportingSkill
- skill-library-generatorSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 1 is distinctive, and 0 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.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 84
Numbers, not adjectives: Never "works" / "ready" — write the test count, the gate output, the measured value.
≈ “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
build-log
Maintain {{build_log_dir}} as a write-once series of numbered entries
(NN_kebab_topic.md) recording phase completions and gate results with their numerical
evidence. The build log is the raw "methods" material for the eventual paper and the
project's replay memory after long gaps — every entry must be worth citing in one of those
two roles.
When to use
The threshold is deliberately high. Open an entry only for:
- Phase completion — all of a phase's modules and tests green, gate passed.
- Fixing a gate result — a correctness-gate tolerance update, a first successful end-to-end run of new machinery, a first result curve.
- Documenting a decision's implementation — the decision itself goes in
{{decision_log}}; the build-log entry carries its execution and numerical proof (comparison tables, figures, before/after numbers).
The practical test: would the paper's methods section, or a future you reconstructing the project, cite this entry? If no — do not open one.
When NOT to use
- A single module was added →
git log+ the module's docstring suffice. - A decision was just written → wait; open the entry when its first implementation + evidence exist.
- Transient exploratory probe → notebook (
project-layout). - Refactor / cleanup / docstring pass → the commit message is the record.
- Anything git already records well (commit messages, PR descriptions).
Numbering and filename
Find the last number: ls {{build_log_dir}} | grep -E '^[0-9]{2}_' | sort | tail -1.
New entry = next number, NN_kebab_topic.md, topic short: module or main deliverable
(e.g. 09_baseline_comparison.md).
Required structure (in order)
# <Module or milestone title>
**Date**: YYYY-MM-DD
**Phase**: <N> (short scope)
**Status**: ✅ / 🟡 + one-sentence summary
**Test total**: N passed, M skipped (from real test-runner output, if applicable)
## Related artifacts
- [path](relative/path) — one-sentence description (evidence dirs, notebooks, decision entries)
## Why this exists <!-- or: Final configuration / Decision rationale -->
Why it was built or measured, which alternative lost, which reference it rests on.
## What was implemented
| File | Public API | Responsibility |
| --- | --- | --- |
| [`path`](relative/path) | `fn(args) -> ret` | one sentence |
## Tests / evidence
- `tests/test_x.py` — N tests, what they pin down.
- Numerical results as tables (sweep tables, before/after), not prose.
## References
- Spec section, predecessor entry, decision numbers, external sources actually used.Rules
- Write-once. Entries are immutable evidence; corrections and follow-ups are a new entry that references the old one. The single allowed retro-edit: marking the previous entry's "next step" pointer as done.
- Numbers, not adjectives. Never "works" / "ready" — write the test count, the gate output, the measured value. A claim without a number or an artifact link is filler.
- Tables over paragraphs — mandatory for implemented-module inventories and numerical results.
- Relative, clickable links to every artifact named — paths a reader can follow from the file's location.
- The why-section carries the rationale — not "added X" but "X solved
<problem>because<reason>, over<alternative>". - Status markers: ✅ completed, 🟡 in progress, ⬜ not started (optionally ⚠ at-risk, ❌ blocked). Do not invent alternatives.
- Entries link their evidence in
{{evidence_dir}}— an entry claiming a run that has no evidence directory is treating "it ran fine" as evidence, which the evidence-convention skill forbids.
Configuration
{{build_log_dir}}— the entry directory (defaultdocs/build_log/).{{decision_log}},{{evidence_dir}}— for cross-links.
Provenance & maintenance
Generalized from a 20+-entry build log in a working computational-research repository
(high-bar entry criteria, header block, artifact list, why-section, implementation table,
write-once discipline); see the pack's examples/ directory for the worked exemplar
mapping. The structure block mirrors the exemplar's entry template with domain content
removed; the "would the paper cite it" test is verbatim exemplar practice.
Re-verify in your project:
ls {{build_log_dir}} | grep -E '^[0-9]{2}_' | sort | tail -3— numbering is contiguous and kebab-cased.grep -L '## References' {{build_log_dir}}/*.md— entries missing a references section (should print nothing).git log --follow --oneline -- {{build_log_dir}}/<oldest entry>— old entries should show no post-creation substantive edits (write-once check).
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.

