engineering-audit
Skill2 starsChanged 7 days agoMITView on GitHub
Audit code with an evidence-first, pipeline-aware review format. Use for engineering audits, systems reviews, hot-path and Big-O analysis, pipeline inspection, resource and budget, trace and identification, product-surface, privacy, performance, observability, security, contract, learned-systems, and lineage reviews. Modes skip when they do not apply. Language- and framework-agnostic. When the system has external services, devices, operators, or physical actuators, also run boundary and control mode. For every executable tree, also run resource and budget mode and trace and identification mode. Skip those two when the tree never becomes executed bytes.
3 other projects use a version of this.
Adopted by
Most recently taken up by
- amkisko/searxng.rbword for word, first seen Nov 1, 2026
- Fmarzochi/EGCtheir own version, first seen Sep 16, 2026
- qwert702/skill-engineering-audittheir own version, first seen Jul 4, 2026
Other versions3 projects use a version of this.
Word for word · 1
- amkisko/searxng.rb★ 4100% same
Edited · 0
None found.
Wrote their own version · 2
- Fmarzochi/EGC★ 61same name
- qwert702/skill-engineering-audit★ 1same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository21 other files it gives its agents.
- AGENTS.mdAGENTS.md
- agent-artifactSkill
- agent-discoverySkill
- agent-run-supervisionSkill
- api-contract-migrationSkill
- changelog-updateSkill
- change-reviewSkill
- ci-reviewSkill
- claims-auditSkill
- dependency-auditSkill
- derivation-auditSkill
- durable-cacheSkill
- infer-conventionsSkill
- operational-signal-intakeSkill
- planned-releaseSkill
- preferred-stackSkill
- pressure-benchmarkingSkill
- public-surface-reconSkill
- rfc-processSkill
- security-auditSkill
- session-reviewSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 2 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
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
- Distinctive: line 18
Use when asked for an engineering audit, systems review, hot-path analysis, Big-O review, pipeline-style inspection, resource and budget …
- Very specific: line 24
When the system has a cache or rebuildable derived state, use durable-cache when that skill is available.
- Distinctive: line 36
When the system has external services, devices, operators, or physical actuators, also run boundary and control mode: who commands whom, …
≈ “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
Engineering audit
Use when asked for an engineering audit, systems review, hot-path analysis, Big-O review, pipeline-style inspection, resource and budget review, trace and identification review, or a product-health pass.
Read engineering-audit.md in this skill directory for dimensions, indicators, stage checks, boundary and control mode, optional product modes, finding format, and ranking. Read resource-and-budget.md and trace-and-identification.md for every tree that can execute. Skip those two files only when the tree never becomes executed bytes, and state that reason.
Read the matching optional file when the product has that surface: product-surface.md, privacy.md, performance.md, observability.md, security.md, contracts.md, learned-systems.md, lineage.md. When security mode runs, also read every applicable companion named in security.md. Skip each file whose surface is absent and state the reason. Stay on concepts. Do not require a named framework or vendor tool.
When the system has a cache or rebuildable derived state, use durable-cache when that skill is available. This skill owns cross-stage evidence, findings, and ranking; durable-cache owns the cache contract and lifecycle. Continue with the cache stage checks here when durable-cache is unavailable.
Quick reference
Pipeline:
ingress -> app logic -> cache -> database -> queue -> worker -> external API -> egressOrder findings by danger, then certainty, then impact, then fix cost. Present the smallest credible fix before structural rewrite. Separate missing coverage from futile coverage. Indicators are categorizable, measurable, and representable.
When the system has external services, devices, operators, or physical actuators, also run boundary and control mode: who commands whom, interface deviations, and whether intended, commanded, reported, inferred, and physical state can diverge without an alarm.
When the product has a place the person can return to, also ask whether unknown identifiers, auth returns, forbidden, error-body status, timeouts, rate limits, 5xx survival of answers, write conflicts, system-initiated recreation, file controls, and deep links, live refresh, or reconnect that replace in-progress work stay honest for the person and for status-based counters.
For every tree that can execute, also run resource and budget mode: measure memory, CPU, storage, network, and energy against a named ceiling; label cheaper, smaller, or greener as inference until benched. Also run trace and identification mode: which emissions identify a person or device, which observers see them, and whether they appear in what this product showed the person for this run. Skip those two modes only when the tree never becomes executed bytes.
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.