review
Skill0 starsChanged 3 months agoMITView on GitHub
The weekly scorecard. What moved, what stalled, what the money did, measured against the vision file, then at most ONE change to how the business works, defended. Run when the user says "how's the week", "score me", "weekly review", "what did we get done", or at the charter's weekly ritual. Solo founders fail by thrash more than by laziness; this skill is the anti-thrash law.
192 other projects use a version of this.
Adopted by
Most recently taken up by
- dzhng/skillstheir own version, first seen Nov 15, 2026
- muxy-app/muxytheir own version, first seen Nov 15, 2026
- macro-inc/macrotheir own version, first seen Nov 15, 2026
- yagizdo/quivertheir own version, first seen Nov 15, 2026
- markmhendrickson/neotomatheir own version, first seen Nov 15, 2026
- felipeam86/pipe-skillstheir own version, first seen Nov 15, 2026
Used with
Projects with it also use close-spec, daily-arxiv 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.
dzhngSkillclose-spec
When a spec under specs/<feature>/ is done shipping (a write-spec build, or any planned task that produced a spec), archive it to specs/done/ and rewrite it from a build-plan into a durable rationale — the why, the principles, the invariants — pointing back to the real code for the how. Use when implementation has landed and the plan no longer matches what shipped, or the user says a feature/spec is finished. Pairs with [write-spec](../write-spec/SKILL.md) (the plan this closes) and [review](../review/SKILL.md) (run before closing).
Also in 5 projects that use this
LambenthanSkilldaily-arxiv
每日从 arXiv 拉取新论文,过滤相关性,auto-ingest 高优先级论文,检测 SOTA 更新
Also in 5 projects that use this
dzhngSkilldesign-with-images
Explore UI and visual design with image generation, then iterate the real implementation against the user-selected concept. Use when brainstorming visual directions, offering mockup variants, or implementing an approved generated design; do not stop at the mockups or the first code approximation.
Also in 5 projects that use this
LambenthanSkillexp-eval
实验判决门:Review LLM 独立评判实验结果 → 4 种判决路径 → 自动更新 claims confidence、ideas status、graph edges
Also in 5 projects that use this
- +5 more from Lambenthan
- +3 more from dzhng
Other versions192 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 192
- garrytan/gstack★ 142ksame name
- ansible/ansible★ 71ksame name
- docling-project/docling★ 71ksame name
- traefik/traefik★ 66ksame name
- ClickHouse/ClickHouse★ 51ksame name
- temporalio/temporal★ 24ksame name
- ccch1mneyyy/dsh-Go★ 24ksame name
- tisfeng/Easydict★ 15ksame name
- juicedata/juicefs★ 15ksame name
- elie222/inbox-zero★ 13ksame name
182 more not listed
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.
The file
We split this skill into the rules it gives an agent
9 rules. 0 are common, 4 are distinctive, and 5 are very specific to this project.
0
Common
Most of the 156 edited versions we read keep these.
4
Distinctive
This author's own choices. Worth a look.
5
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 29
Read references/CONVENTIONS-core.md and run its Universal preflight.
- Very specific: line 42
Read that inclusive period slice, citing everything: log.md rows inside the boundaries, baton.md and each active initiative's Log (the …
- Distinctive: line 53
Every claim in the scorecard traces to a row or a file.
- Distinctive: line 60
When a downstream note repeats a claim, check the original receipt instead of trusting the note.
- Distinctive: line 70
If the founder reports another post but supplies no text, record only the bounded publication claim as founder-stated/unverified, ask for …
- Very specific: line 74
Before moving content to approved or published, capture the founder's exact approval or publication words in an immutable …
- Distinctive: line 123
Ask before you assert: never present a quantity, duration, sample size, date, or causal motive in prose before asking the founder for it or …
- Very specific: line 140
Write the review note to wiki/business/reviews/review-<period_end>.md with base-five frontmatter plus type: review, period_start, …
- Very specific: line 171
Run scripts/graph-audit on every touched durable file, including every created or reconstructed content artifact.
“Common” is measured: kept by at least 6 in 10 of the 156 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.
Files
review: honest scorecard, one change, back to work
Diagnose freely, change the machine once. The founder who rebuilds their system every week ships systems instead of shipping the business. Review makes the week legible, then enforces the discipline that makes next week better instead of merely different.
the week
|
[ evidence ] log rows, learnings, decisions, roadmap movement
|
[ scorecard ] vs wiki/business/vision.md, failures verbatim
|
[ ONE change ] at most one, defended, on the recordStep 0: Preflight
Read references/CONVENTIONS-core.md and run its Universal preflight. A Fresh,
Partial, or Wrong-folder result stops and runs the co-founder-setup skill before anything else (re-sync for Partial), without writing a review. On Ready, read the charter and path map.
Done when: the founder system is Ready and the review evidence paths are known.
Step 1: Gather the evidence
Set the review period before scoring anything. Find the latest type: review page in
wiki/business/reviews/. For the first review, period_start is the date of the genesis
setup row in log.md; later reviews start the day after the latest period_end.
period_end is today. If the latest review already ends today, update that same review only
after explicit confirmation; never create an overlapping period.
Read that inclusive period slice, citing everything: log.md rows inside the boundaries, baton.md
and each active initiative's Log (the canonical record of what was COMMITTED: the named next
moves and dates that define "supposed to move"), learning notes written (with their
cost/result), decision rows, initiative movement (steps checked, states flipped, fog
graduated), anything still sitting in "Needs your call", and the previous review note's
"Next week's candidate" section, carried back as a candidate only, never as pre-approved
work. Read wiki/business/metrics.md as the financial source of record; each number must have a
dated source. Learning cost/result fields may explain a metric but never silently replace a
missing total. The vision file is the bar; no vision file means the review runs shallow and says so,
recommending the vision skill first.
Every claim in the scorecard traces to a row or a file. A week with thin logs gets reported as thin on record, never reconstructed from vibes.
Apply the shared durable claim admission gate while gathering, not after scoring. Reopen every
cited file and quote the row that supports the claim. A one-row tracker proves one logged block,
not a 14-day measurement. Monthly recurring revenue is a run-rate, not weekly cash received or
weekly net. A founder-reported result with no instrument row is Founder-stated: unverified.
When a downstream note repeats a claim, check the original receipt instead of trusting the note.
Reconcile lifecycle evidence before scoring it. If the founder says a piece shipped, confirm the
date, update its canonical content_state and published_at, then reconcile its index/hub,
serving roadmap checkbox, publication metric, and eligible voice-sample reference. Until that
transaction is complete, report the statement as founder-stated/unverified and do not score the
piece as shipped. Apply the same rule to customer evidence and roadmap completion conditions.
Review never authors or reconstructs missing customer-facing copy. A reported publication can advance an existing content file only when the vault already contains the exact published text. If the founder reports another post but supplies no text, record only the bounded publication claim as founder-stated/unverified, ask for the exact copy, and leave content lifecycle plus roadmap unchecked. Content-engine owns any new draft.
Before moving content to approved or published, capture the founder's exact approval or
publication words in an immutable raw/inbox/session-receipt-<date>-<slug>.md. Set
approval_receipt: <path>#<exact words> and, for publication,
publication_receipt: <path>#<exact words>. An edit to one phrase is not whole-artifact approval.
Graph-audit rejects lifecycle state without these checked anchors. These are T1 consent surfaces
(shared contract, Immutable consent receipts): with no founder yes on record, review does not
advance the content state and does not write the approval field.
Done when: the week's facts are collected with their sources.
Step 2: The scorecard
One screen, against the vision's quarter goal and direction:
- Moved: what actually advanced, with the receipt.
- Stalled: what was supposed to move and did not. Verbatim, unlaundered: "the pricing page did not happen" is a fact; "made progress on positioning" covering the same week is a lie with better manners.
- The money: what came in, what went out, what it says (from the period's dated rows in
wiki/business/metrics.md, with learning cost/result as explanation; "not tracked for this period" is a legitimate, flagged entry). - The vault: lessons banked this week, and notes now past their
review_afterdate (flag the stale list for steward's next health-check). - Score vs the quarter goal: on track / behind / off, in one line with the number that says so.
Done when: the founder sees the week as it was, not as it felt.
Step 3: The one change
At most ONE change to how the business works this week: a system, a habit, a cadence, an offer's mechanics. The founder proposes or picks from your recommendation; either way it gets defended out loud: what problem this change solves, the evidence from the scorecard, what would prove it worked by next review. Diagnosis is unlimited; surgery is rationed.
An internal, reversible workflow adjustment cites the scorecard evidence that motivates it; it uses the shared reversible-internal-experiment contract: founder yes, review date, success condition, and undo condition. It does not manufacture an outside source. A customer-facing, money, strategy, readiness-state, or hard-to-reverse change is consequential and clears both receipt legs before it can become the week's one change. A missing leg parks the candidate in the shared canonical shape; it is not the chosen change.
The founder's explicit yes is a precondition, not prose review may supply. The Founder's yes:
field is a T1 consent surface (shared contract, Immutable consent receipts): capture the exact yes
in the immutable raw/inbox/session-receipt-<date>-<slug>.md and link it as
[receipt: raw/inbox/<file>#<exact words>] before writing the one change or any decision row.
Without that anchor, leave the candidate unchosen and write no Founder's yes: field; a supplied
date is never a substitute for the words.
Ask before you assert: never present a quantity, duration, sample size, date, or causal motive in prose before asking the founder for it or citing a receipt for it.
Apply the shared parameter gate before making the change concrete. A duration, cadence, or target
comes from an admitted receipt or one founder question. Without it, recommend the behavior without
an invented number or preserve Working hypothesis: unverified as next week's candidate.
The urge for a second change goes on the record under "next week's candidate" and waits. A change that is hard to reverse, surprising, and a real tradeoff earns a decision row.
Zero changes is a legitimate outcome; a machine that worked this week does not need improving to feel productive.
Done when: at most one change is on the record with its defense, or none, said plainly.
Step 4: Close the loop
Write the review note to wiki/business/reviews/review-<period_end>.md with base-five
frontmatter plus type: review, period_start, period_end, and review_state: complete.
List it in index.md under Reviews and link hub/business.md. The note's required sections:
the scorecard, the one change and its defense,
"Next week's candidate", "Steward queue" (notes past review_after, or "None."), and
"Bank queue" (reviewed initiatives and the lesson each should bank, or "None.").
For every non-None item, append a pending entry to queue.md: review → steward for stale
notes, review → bank for reviewed initiatives. Put the queue IDs in these note sections.
Queue items that live only in chat or inside a review paragraph are not queue items. Append
## [period_end] review | <period_start> to <period_end> to log.md.
state: reviewed has a checkable bar: the current state is exactly executing, every box under
the initiative's Next steps is checked OR the remaining unchecked work has been moved to
Roadmap → Stopped with a founder-confirmed date and reason, the scorecard names its outcome this week, "Needs your
call" is "None." or explicitly non-blocking, and Parked recommendations is "None." Re-read the
state immediately before writing it. planned work routes to sprint or state repair; idea and
survived route to their missing gate; reviewed and banked remain closed. Initiatives
passing the legal executing → reviewed transition get the state flip plus a Log line:
- [date] review: outcome judged; hand to bank for <lesson topic>.
Move each changed initiative to the matching hub bucket and update the index summary, baton,
roadmap conditions, and queue projection in the same linked-state transaction.
Quarterly cadence, deterministic: 12 review rows in log.md since the last vision row,
or 84+ days since vision.md was updated, adds a "Vision nudge" line to the note and the
close; the quarter goal just got scored 12 times, and the founder knows things they did not
know when they set it.
Close in chat: the score line, the one change (or none), the ONE next move.
Mandatory filing gate
- Run
scripts/graph-auditon every touched durable file, including every created or reconstructed content artifact. - Fix every violation and rerun it. The review, founder approval, lifecycle changes, and queue writes remain incomplete until it exits zero.
Done when: the note is filed, the log row exists, and next week's review can measure this week's change.
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.

