freeze-readiness-check
Skill1 starsChanged 4 months agoMITView on GitHub
Judge whether the current requirement/demo/PRD work is stable enough to freeze as a review or delivery baseline, and identify what blocks freezing when it is not ready. Use when a product workflow round may be turned into a versioned snapshot, when Codex needs to decide whether to keep iterating or preserve the current baseline, or when downstream embedded-PRD/project-ops work depends on a trustworthy freeze decision.
In this repository17 other files it gives its agents.
- audience-judgment-framingSkill
- embedded-prd-auditSkill
- memory-export-summarySkill
- pm-demo-designSkill
- pm-design-absorptionSkill
- pm-devil-advocateSkill
- pm-embedded-prdSkill
- pm-prd-writerSkill
- pm-project-opsSkill
- pm-react-prototype-executionSkill
- pm-requirement-intakeSkill
- pm-requirement-reverse-auditSkill
- pm-value-anchorSkill
- prototype-export-intakeSkill
- request-triageSkill
- rule-change-signalSkill
- state-auditSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
33 rules. About 0 are common, 32 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈32
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 this skill to stop teams from freezing too early because a demo looks finished or a document feels long enough.
- Distinctive: line 34
If the input is too incomplete to judge honestly, say that clearly and ask for the smallest missing piece.
- Distinctive: line 47
Do not require every artifact to exist before starting judgment.
- Distinctive: line 51
Always judge readiness relative to one of these targets.
- Distinctive: line 55
Use when the goal is:
- Distinctive: line 57
keep a stable discussion baseline
- Distinctive: line 58
avoid losing the exact round under review
- Distinctive: line 65
treat the round as a trustworthy downstream baseline
- Distinctive: line 74
If the target is not stated, infer the most likely one:
- Distinctive: line 75
if the round is mainly for review/discussion → review freeze
- Distinctive: line 76
if the round is meant to act as a trusted continuation baseline → delivery freeze
- Distinctive: line 82
Ask whether the requirement side is stable enough for the target.
- Distinctive: line 99
Ask whether the demo side is stable enough for the target.
- Distinctive: line 115
If embedded PRD / linked-delivery work is in scope, ask:
- Distinctive: line 117
do doc and demo belong to the same effective round?
- Distinctive: line 118
do current mappings still tell the truth?
- Distinctive: line 129
if this freezes now, can later work tell what it replaced and what it contains?
- Distinctive: line 147
Not a reason to stop freeze, but should be carried forward visibly.
- Distinctive: line 152
not ready to freeze
- Distinctive: line 157
Do not flatten these into only yes/no.
- Distinctive: line 161
Prefer this compact structure:
- Distinctive: line 177
Keep it compact, but do not omit:
- Distinctive: line 185
Use these tests when the judgment is borderline.
- Distinctive: line 189
If this froze today, would people treat unstable logic as stable truth?
- Distinctive: line 191
If yes, do not freeze at that level.
- Distinctive: line 197
If yes, do not approve delivery freeze.
- Very specific: line 201
Do rule/demo/PRD/mapping assets still describe the same effective round?
- Distinctive: line 203
If no, block stronger freeze.
- Distinctive: line 209
If no, avoid freeze or downgrade to a weaker freeze target.
- Distinctive: line 217
no major deception is created as long as instability is labeled
- Distinctive: line 253
Read these only when useful:
- Distinctive: line 260
When freeze is recommended:
- Distinctive: line 265
When freeze is blocked:
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 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
Freeze Readiness Check
Act as a command-style skill for deciding whether a current PM workflow round is ready to freeze.
Core job
Evaluate the current round across requirement stability, demo stability, linked-delivery stability, and versioning coherence, then output:
- whether freeze is recommended
- what kind of freeze is justified, if any
- what the main blockers are
- what should happen next
Use this skill to stop teams from freezing too early because a demo looks finished or a document feels long enough.
Boundaries
This skill is responsible for:
- judging freeze readiness
- identifying blockers and risk level
- distinguishing review-candidate freeze from stronger delivery freeze
- telling upstream/downstream what should happen next
This skill is not responsible for:
- rebuilding the requirement from scratch
- rewriting the demo or PRD itself
- performing version management actions directly
- deciding final business truth when major requirement disputes are still open
If the input is too incomplete to judge honestly, say that clearly and ask for the smallest missing piece.
Inputs
Work from whatever current-round evidence exists, such as:
- requirement skeleton or rule baseline
- demo status
- embedded-PRD or linked-delivery status
- unresolved items / unfrozen items
- recent rule-change signals
- intended freeze target (
reviewor stronger delivery/frozen baseline) - version/snapshot context if available
Do not require every artifact to exist before starting judgment. Judge based on the claimed freeze target.
Freeze targets
Always judge readiness relative to one of these targets.
1) Review freeze
Use when the goal is:
- preserve a coherent review candidate
- keep a stable discussion baseline
- avoid losing the exact round under review
This target tolerates some unfrozen items, as long as they are clearly marked and do not destroy coherence.
2) Delivery freeze
Use when the goal is:
- treat the round as a trustworthy downstream baseline
- hand work into embedded PRD / project ops / later continuation with minimal reinterpretation
This target requires stronger rule closure, stronger state closure, and stronger cross-artifact alignment.
Working flow
1) Identify the claimed freeze target
If the target is not stated, infer the most likely one:
- if the round is mainly for review/discussion → review freeze
- if the round is meant to act as a trusted continuation baseline → delivery freeze
State the inferred target if you had to infer it.
2) Check requirement stability
Ask whether the requirement side is stable enough for the target.
Check:
- goal and scope closure
- main rule stability
- conflict/priority stability
- special-trigger stability
- config/history logic clarity
- whether the rule model is converged or still example-pile driven
Typical outcomes:
- ready
- borderline
- blocked
3) Check demo stability
Ask whether the demo side is stable enough for the target.
Check:
- key pages/modules/states are covered
- missing states are minor vs major
- current demo truthfully expresses current rules
- exploratory branches are separated from baseline branches
- recent changes are presentation-level vs rule-level
Typical outcomes:
- ready
- borderline
- blocked
4) Check linked-delivery stability
If embedded PRD / linked-delivery work is in scope, ask:
- can stable vs unfrozen items be separated cleanly?
- do doc and demo belong to the same effective round?
- do current mappings still tell the truth?
- would a reviewer/downstream consumer be misled by the current assembly?
For review freeze, partial linked-delivery incompleteness may be acceptable. For delivery freeze, misleading linkage is a strong blocker.
5) Check versioning coherence
Ask:
- is there a clear version/stage identity for this round?
- is the current baseline distinguishable from working experiments?
- if this freezes now, can later work tell what it replaced and what it contains?
This does not require final project-ops execution yet, but the round should be coherent enough to preserve.
6) Judge blockers
Classify blockers as:
- hard blocker
- soft blocker
- watch item
Hard blocker
Freeze would create false truth or mislead downstream work.
Soft blocker
Freeze is possible, but there is still obvious quality debt or ambiguity worth fixing first.
Watch item
Not a reason to stop freeze, but should be carried forward visibly.
7) Output freeze recommendation
Choose one:
- not ready to freeze
- ready for review freeze only
- ready for freeze with explicit caveats
- ready for delivery freeze
Do not flatten these into only yes/no.
Output format
Prefer this compact structure:
## Freeze readiness
- Freeze target:
- Recommendation:
- Requirement stability:
- Demo stability:
- Linked-delivery stability:
- Versioning coherence:
- Hard blockers:
- Soft blockers:
- Watch items:
- Recommended next move:Keep it compact, but do not omit:
- target
- recommendation
- blockers
- next move
Decision heuristics
Use these tests when the judgment is borderline.
Test 1: Truthfulness test
Ask:
- If this froze today, would people treat unstable logic as stable truth?
If yes, do not freeze at that level.
Test 2: Misleading-baseline test
Ask:
- Would downstream embedded-PRD, project-ops, or later continuation inherit a misleading baseline?
If yes, do not approve delivery freeze.
Test 3: Current-round coherence test
Ask:
- Do rule/demo/PRD/mapping assets still describe the same effective round?
If no, block stronger freeze.
Test 4: Unfrozen-item separability test
Ask:
- Can unsettled items be cleanly marked and kept from contaminating stable baseline claims?
If no, avoid freeze or downgrade to a weaker freeze target.
Typical recommendations
Ready for review freeze only
Use when:
- the round is coherent enough to preserve and discuss
- some items remain unfrozen
- no major deception is created as long as instability is labeled
Ready for freeze with explicit caveats
Use when:
- most of the round is stable
- a few non-core blockers remain
- preservation is still useful and the caveats can be carried clearly
Ready for delivery freeze
Use when:
- rule model is stable
- key states and outcomes are stable
- demo truthfully expresses the current baseline
- linked-delivery assembly is truthful
- versioning handoff can be performed without ambiguity
Not ready to freeze
Use when:
- major rule-change signals remain unresolved
- key states are missing
- demo is still exploratory in core areas
- doc/demo/mapping no longer align
- baseline truth would be overstated by freezing now
Anti-patterns
Avoid these failures:
- freezing because the UI looks polished
- freezing because people are tired of discussing
- confusing review freeze with delivery freeze
- freezing a mixed bundle of stable and unstable work with no labels
- treating a working branch as if it were already the current baseline
- ignoring unresolved Class 3 rule-change signals
References to load when needed
Read these only when useful:
../pm-requirement-intake/references/readiness-gates.mdfor upstream readiness logic and rule-model convergence thinking../pm-project-ops/references/version-and-snapshot-rules.mdfor stage labels, baseline sets, and freeze/snapshot discipline../pm-demo-design/references/rule-change-signal-template.mdwhen unresolved rule-change signals are part of the block
Handoff expectations
When freeze is recommended:
- say what level of freeze is justified
- say what caveats must stay visible
- make clear whether downstream project-ops should preserve a review baseline or a stronger delivery baseline
When freeze is blocked:
- say whether the next move belongs mainly in requirement intake, demo design, embedded PRD cleanup, or versioning cleanup
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.

