Skip to content

Commit 05d573a

Browse files
committed
1.5.1 Tighten test and workflows
1 parent b15f450 commit 05d573a

16 files changed

Lines changed: 114 additions & 66 deletions

File tree

‎AGENTS.md‎

Lines changed: 10 additions & 8 deletions
Original file line numberDiff line numberDiff line change
@@ -123,21 +123,23 @@
123123
changed observable contracts, reported regressions, meaningful boundaries,
124124
and plausible costly failures, especially security, trust-boundary, or
125125
data-loss risks.
126-
- Use one representative case per equivalent behavior class. Skip redundant
127-
permutations, implementation-detail assertions, and contrived or unreachable
128-
states unless a requirement or past defect justifies them. Do not introduce a
126+
- Use one representative case per necessary observable contract or reported
127+
regression. Skip speculative edge cases, redundant permutations,
128+
implementation-detail assertions, and contrived or unreachable states.
129+
Retain boundary coverage only for an explicit requirement, past defect, or
130+
plausible security, trust-boundary, or data-loss failure. Do not introduce a
129131
test framework solely to satisfy this rule; if automated coverage is
130132
impractical, explain why and perform the strongest targeted verification.
131133
- Treat existing tests as regression contracts. Preserve their assertions unless
132134
the requested behavior intentionally changes. When behavior changes, update
133135
only the affected tests and add coverage for the new contract; never weaken or
134136
delete tests merely to make the suite pass.
135-
- Run the smallest targeted verification that meaningfully validates a change
136-
before the full required checks. Avoid broad commands and use the repository's
137-
documented package manager and scripts.
137+
- Run the smallest targeted verification that meaningfully validates a change.
138+
Broaden validation only for affected shared behavior, a failure, or an explicit
139+
release requirement; a command being available or fast is not enough.
138140
- Do not change dependencies, global tools, or the environment by default.
139-
- Run `pnpm run typecheck`, `pnpm test`, and `pnpm run pack:check` before
140-
declaring a change complete.
141+
- Run `pnpm run typecheck`, `pnpm test`, and `pnpm run pack:check` for release
142+
verification and package-wide changes that can affect those contracts.
141143

142144
## Template and release workflow
143145

‎MAINTAINERS.md‎

Lines changed: 9 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -100,6 +100,7 @@ pnpm run pack:check
100100
Inspect the dry-run contents. They should contain only:
101101

102102
- `assets/agents/*.toml`
103+
- `assets/skills/codex-kit-audit-agents/**`
103104
- `assets/skills/codex-kit-reconcile-agents/**`
104105
- `assets/SUBAGENT_ROUTING.md`
105106
- `assets/TEMPLATE_AGENTS.md`
@@ -255,11 +256,14 @@ reconciliation skill without duplicating its procedure. Downstream users must up
255256
refresh the global installation before syncing to receive both the enhanced
256257
skill and template.
257258

258-
`project audit` is separate and read-only: it validates only the target
259-
directory, then prints a marked prompt invoking `$codex-kit-audit-agents`. It
260-
must not contact npm, inspect project state, register the project, or write any
261-
file. Users may run it deliberately after substantial instruction changes or on
262-
an optional biweekly/monthly cadence; it is never an always-on workflow.
259+
`project audit` is separate from synchronization: the CLI validates only the
260+
target directory, then prints a marked prompt invoking
261+
`$codex-kit-audit-agents`. The skill audits instruction, skill, workflow, and
262+
test scope, applies only unambiguous cleanup, and preserves justified regression
263+
and costly-risk coverage. The CLI must not contact npm, inspect project state,
264+
register the project, or write any file. Users may run it deliberately after
265+
substantial instruction changes or on an optional biweekly/monthly cadence; it
266+
is never an always-on workflow.
263267

264268
## Verification
265269

‎README.md‎

Lines changed: 8 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -184,11 +184,14 @@ codex-kit project mark-applied
184184
modify `AGENTS.md`. Use `codex-kit project status` to check whether the current
185185
template still needs reconciliation.
186186

187-
Run `codex-kit project audit` whenever you want a standalone instruction review
188-
(monthly or biweekly is a reasonable optional cadence). It only validates the
189-
target directory and prints a marked prompt invoking the explicit-only
190-
`$codex-kit-audit-agents` skill. The CLI does not contact npm, require project
191-
initialization, inspect or modify state, register the project, or write files.
187+
Run `codex-kit project audit` whenever you want a standalone review of project
188+
instructions, skill/workflow validation scope, and clearly redundant or
189+
speculative tests (monthly or biweekly is a reasonable optional cadence). The
190+
audit preserves justified regression and costly-risk coverage and reports
191+
ambiguous reductions. The CLI itself only validates the target directory and
192+
prints a marked prompt invoking the explicit-only `$codex-kit-audit-agents`
193+
skill; it does not contact npm, require project initialization, inspect or
194+
modify state, register the project, or write files.
192195

193196
## Options
194197

‎RELEASE_NOTES.md‎

Lines changed: 3 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -1,6 +1,3 @@
1-
- Add task-scoped downstream guidance that keeps universal safeguards in `AGENTS.md` and loads detailed project skills and references only when relevant.
2-
- Expand reconciliation and initialization audits to classify existing and incoming guidance, preserve local adaptations, and validate narrow workflow routing.
3-
- Shorten project prompts so the packaged reconciliation skill owns semantic restructuring, with updated upgrade and global-refresh guidance.
4-
- Add read-only `project audit` prompting and an explicit-only audit skill managed by global install, list, and uninstall.
5-
- Update Vitest to `4.1.11` to resolve its development-server security advisory.
6-
- Automatically validate, tag, publish, and create the GitHub Release for release-ready pushes to `main`.
1+
- Scope UI/style gates to changes in appearance or interaction, so data, content, event, and backend-only work does not trigger them.
2+
- Make agents run the smallest validation that proves changed behavior and reserve broader suites for shared impact, failures, or release requirements.
3+
- Extend reconciliation and audits to conservatively remove demonstrably redundant or speculative tests while preserving regression and costly-risk coverage.

‎TEMPLATE_AGENTS.md‎

Lines changed: 11 additions & 7 deletions
Original file line numberDiff line numberDiff line change
@@ -138,7 +138,7 @@ triggered skills and selectively read references.
138138
Do not assume tools from other projects.
139139
- Keep discovered stack-specific guidance in the project's
140140
`# Project-Specific Instructions`, not in this shared template.
141-
- For user-facing work, perform a UI/style preflight before editing: inspect the
141+
- When UI appearance or interaction changes, perform a UI/style preflight before editing: inspect the
142142
closest same-purpose shipped features (especially tables, filters, search,
143143
forms, dialogs, and page layouts) and identify their design-system primitives,
144144
tokens, spacing, typography, responsive behavior, interactions, states, and
@@ -168,16 +168,20 @@ triggered skills and selectively read references.
168168

169169
## Commands And Verification
170170

171-
- Avoid broad commands. After changes, run the smallest targeted verification
172-
that meaningfully validates them when practical, then report the command and
173-
result. Use the repository's documented package manager and scripts.
171+
- Select validation from the changed observable behavior and concrete failure
172+
risk. Run the smallest existing check that establishes correctness, then
173+
report the command and result. Run broader suites only when affected shared
174+
behavior, a failure, or an explicit release requirement justifies them; a
175+
command being available, fast, or mentioned in loaded guidance is not enough.
174176
- Select tests for regression value rather than exhaustive coverage. When the
175177
repository has an established test setup, cover changed observable contracts,
176178
reported regressions, meaningful boundaries, and plausible costly failures,
177179
especially security, trust-boundary, or data-loss risks.
178-
- Use one representative case per equivalent behavior class. Skip redundant
179-
permutations, implementation-detail assertions, and contrived or unreachable
180-
states unless a requirement or past defect justifies them.
180+
- Use one representative case per necessary observable contract or reported
181+
regression. Skip speculative edge cases, redundant permutations,
182+
implementation-detail assertions, and contrived or unreachable states.
183+
Retain boundary coverage only for an explicit requirement, past defect, or
184+
plausible security, trust-boundary, or data-loss failure.
181185
- Treat existing tests as regression contracts. Preserve their assertions unless
182186
the requested behavior intentionally changes. When behavior changes, update
183187
only the affected tests and add coverage for the new contract; never weaken or

‎assets/SUBAGENT_ROUTING.md‎

Lines changed: 2 additions & 2 deletions
Original file line numberDiff line numberDiff line change
@@ -22,7 +22,7 @@ trigger `implementer`. Supporting `code-explorer` discovery and the mandatory
2222
`code-reviewer` handoff remain allowed, while implementation subagents require
2323
an explicit user request.
2424

25-
For user-facing work, the root or assigned worker performs a UI/style preflight
25+
When UI appearance or interaction changes, the root or assigned worker performs a UI/style preflight
2626
before editing: inspect the closest same-purpose shipped features (especially
2727
tables, filters, search, forms, dialogs, and page layouts), identify their
2828
design-system primitives, tokens, spacing, typography, responsive behavior,
@@ -37,7 +37,7 @@ screenshot tooling exists, compare the rendered feature with its analogue;
3737
otherwise report rendered comparison unavailable.
3838

3939
Every new or materially changed feature follows this mandatory workflow:
40-
UI/style preflight; implement and stabilize; map each responsibility to its final file; extract
40+
when applicable, UI/style preflight; implement and stabilize; map each responsibility to its final file; extract
4141
independently understandable concerns; validate the decomposed implementation;
4242
then hand off to `code-reviewer`. Pages, routes, controllers, commands, and
4343
entrypoints contain composition and orchestration only. Web page files may keep

‎assets/TEMPLATE_AGENTS.md‎

Lines changed: 11 additions & 7 deletions
Original file line numberDiff line numberDiff line change
@@ -138,7 +138,7 @@ triggered skills and selectively read references.
138138
Do not assume tools from other projects.
139139
- Keep discovered stack-specific guidance in the project's
140140
`# Project-Specific Instructions`, not in this shared template.
141-
- For user-facing work, perform a UI/style preflight before editing: inspect the
141+
- When UI appearance or interaction changes, perform a UI/style preflight before editing: inspect the
142142
closest same-purpose shipped features (especially tables, filters, search,
143143
forms, dialogs, and page layouts) and identify their design-system primitives,
144144
tokens, spacing, typography, responsive behavior, interactions, states, and
@@ -168,16 +168,20 @@ triggered skills and selectively read references.
168168

169169
## Commands And Verification
170170

171-
- Avoid broad commands. After changes, run the smallest targeted verification
172-
that meaningfully validates them when practical, then report the command and
173-
result. Use the repository's documented package manager and scripts.
171+
- Select validation from the changed observable behavior and concrete failure
172+
risk. Run the smallest existing check that establishes correctness, then
173+
report the command and result. Run broader suites only when affected shared
174+
behavior, a failure, or an explicit release requirement justifies them; a
175+
command being available, fast, or mentioned in loaded guidance is not enough.
174176
- Select tests for regression value rather than exhaustive coverage. When the
175177
repository has an established test setup, cover changed observable contracts,
176178
reported regressions, meaningful boundaries, and plausible costly failures,
177179
especially security, trust-boundary, or data-loss risks.
178-
- Use one representative case per equivalent behavior class. Skip redundant
179-
permutations, implementation-detail assertions, and contrived or unreachable
180-
states unless a requirement or past defect justifies them.
180+
- Use one representative case per necessary observable contract or reported
181+
regression. Skip speculative edge cases, redundant permutations,
182+
implementation-detail assertions, and contrived or unreachable states.
183+
Retain boundary coverage only for an explicit requirement, past defect, or
184+
plausible security, trust-boundary, or data-loss failure.
181185
- Treat existing tests as regression contracts. Preserve their assertions unless
182186
the requested behavior intentionally changes. When behavior changes, update
183187
only the affected tests and add coverage for the new contract; never weaken or

‎assets/agents/code-explorer.toml‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -14,7 +14,7 @@ You locate and distill the code relevant to a task so the orchestrator never has
1414
1. Clarify the target — Restate (to yourself) what the caller needs: which behavior, symbol, flow, or convention.
1515
2. Search wide, read narrow — Use grep/glob to fan out, then read only the excerpts needed to confirm relevance. Prefer reading specific line ranges over whole files.
1616
3. Trace the contract — For each relevant piece, note its inputs/outputs, callers, and any invariants or guards the implementer must respect.
17-
4. UI/style preflight — For user-facing work, locate the closest same-purpose shipped UI and report its reusable components, design tokens, layout, responsive behavior, interaction/state, and accessibility conventions. Flag conflicting or missing precedents; do not invent a standard.
17+
4. UI/style preflight — When UI appearance or interaction changes, locate the closest same-purpose shipped UI and report its reusable components, design tokens, layout, responsive behavior, interaction/state, and accessibility conventions. Data, content, event, or backend-only changes do not trigger this step unless they change that UI contract. Flag conflicting or missing precedents; do not invent a standard.
1818
5. Report — Return the structured report below. Nothing else.
1919
2020
## Report format (strict)

‎assets/agents/code-reviewer.toml‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -27,7 +27,7 @@ when inseparable from a file's single responsibility. Allow an exception only
2727
for a concrete framework or tooling constraint explicitly identified in the
2828
handoff; otherwise end with `REQUEST_CHANGES` for violations.
2929
30-
For user-facing changes, review consistency against the closest same-purpose
30+
For changes to UI appearance or interaction, review consistency against the closest same-purpose
3131
shipped UI and applicable written `AGENTS.md`/`PLANS.md` guidance. Check reuse of
3232
existing components, tokens, layout, responsive behavior, interactions, states,
3333
and accessibility conventions. On repository evidence, return

‎assets/agents/implementer.toml‎

Lines changed: 6 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -1,25 +1,25 @@
11
name = "implementer"
2-
description = "Implements features and bug fixes with accompanying unit tests, then validates them by running the test suite. Use to write code and prove it works. Hands the diff back for review; does not self-approve."
2+
description = "Implements features and bug fixes with focused tests when needed, then runs the smallest relevant validation. Use to write code and prove it works. Hands the diff back for review; does not self-approve."
33
model = "gpt-5.6-luna"
44
model_reasoning_effort = "high"
55
developer_instructions = """
66
# Implementer (Code + Unit Tests + Validation)
77
88
Begin your first user-visible response with this progress message exactly once: `Delegating to custom implementer — gpt-5.6-luna, high reasoning.`
99
10-
You write the actual code, its unit tests, and you run those tests to prove the change works.
10+
You write the actual code and run the smallest relevant validation that proves the change works.
1111
1212
## Workflow
1313
1. Understand — Read the relevant files and the assigned slice before editing. Don't guess at contracts.
14-
2. UI/style preflight — Before editing user-facing code, inspect the closest same-purpose shipped features (especially tables, filters, search, forms, dialogs, and page layouts), then record the applicable components, tokens, layout, responsive behavior, interactions, states, and accessibility conventions. Reuse them. If no written standard exists but trustworthy repeated precedent does, derive concise always-on guidance in `AGENTS.md` and feature-specific decisions or exceptions in `PLANS.md`. Pause and ask the user whether to keep, update, or override before divergence, changing a guideline, resolving conflicts, or proceeding without a trustworthy analogue. Use browser or screenshot tooling for rendered comparison when available; otherwise report unavailable.
14+
2. UI/style preflight — When UI appearance or interaction changes, inspect the closest same-purpose shipped features (especially tables, filters, search, forms, dialogs, and page layouts), then record the applicable components, tokens, layout, responsive behavior, interactions, states, and accessibility conventions. Data, content, event, or backend-only changes do not trigger this step unless they change that UI contract. Reuse applicable patterns. If no written standard exists but trustworthy repeated precedent does, derive concise always-on guidance in `AGENTS.md` and feature-specific decisions or exceptions in `PLANS.md`. Pause and ask the user whether to keep, update, or override before divergence, changing a guideline, resolving conflicts, or proceeding without a trustworthy analogue. Use browser or screenshot tooling for rendered comparison when available; otherwise report unavailable.
1515
3. Implement — Make the smallest change that satisfies the slice; follow the discovered UI and code patterns, naming, and idioms. Use secure-by-default patterns: never hardcode secrets — read from env vars / a secrets vault, and leave a `// TODO: load from env or secrets vault` marker where a credential belongs.
16-
4. Unit test — Add or update unit tests covering new behavior, edge cases, and failure paths.
16+
4. Unit test — Add or update one representative test for each necessary observable contract or reported regression. Keep boundary coverage only for an explicit requirement, past defect, or plausible security, trust-boundary, or data-loss failure; skip speculative edge cases and redundant permutations.
1717
5. Semantic decomposition — Map each responsibility to its final file, then extract independently understandable concerns before validation. Pages, routes, controllers, commands, and entrypoints orchestrate only; page files do not define child components, substantial UI sections, or domain logic. Put independently changeable UI in descriptive feature-local files, and move independently testable hooks, schemas, data access, transformations, and domain logic out of presentation. Avoid generic dumping grounds; keep components feature-local unless shared/global by evidence. Tiny inseparable helpers or markup may remain inline. An unrelated small fix need not trigger broad refactoring. Name any concrete framework/tooling exception in the handoff.
18-
6. Validate — Run tests against the decomposed implementation (and linters/build if quick). Paste the actual result. If tests fail, fix and re-run — never report success on red.
18+
6. Validate — Run the smallest existing check that proves the changed observable behavior. Run broader tests, lint, formatting, or builds only when affected shared behavior, a failure, the changed files, or an explicit release requirement justifies them. Paste the actual result. If validation fails, fix and re-run — never report success on red.
1919
7. Hand off — Send every completed feature to `code-reviewer` for structural and UI consistency review. Report the responsibility map, extracted files, applicable precedent/standard, tests, rendered comparison or its unavailability, and output. State plainly if anything is unverified.
2020
2121
## Rules
22-
- No change is done until its tests are green and you've shown the output.
22+
- No change is done until its necessary validation is green and you've shown the output.
2323
- Finish within five minutes. At the deadline, stop running commands and return the stable edits with exact incomplete or failing validation evidence.
2424
- Run one validation process at a time. Stop a command that produces no result within two minutes unless repository guidance documents a longer normal runtime; report it as hung instead of repeatedly retrying it.
2525
- Don't expand scope beyond the assigned slice; flag anything else you notice.

0 commit comments

Comments
 (0)