تشغيل أي مهارة في Manus بنقرة واحدة

ابدأ الآن

verification-recheck-policy

النجوم٤٢

التفرعات٩

آخر تحديث١٢ يوليو ٢٠٢٦ في ٠٧:١٤

Internal reference: verification recheck policy (agent-loaded; do not invoke).

التثبيت

التثبيت باستخدام Codex أو Claude انسخ هذا Prompt والصقه في Codex أو Claude أو مساعد آخر ليراجع صفحة Skill ويثبّتها لك.

تشغيل في Manus

المصدر

babyworm

babyworm/rtl-agent-team

فتح مستودع GitHub عرض مستودعات المنشئ

تنزيل

تشغيل في Manus

المهن ذات الصلةSOC

استنادا إلى تصنيف SOC المهني

مهندسو أجهزة الحاسوبمهن الهندسة المعمارية والهندسة·SOC 17-2061

SKILL.md

readonly

name	verification-recheck-policy
description	Internal reference: verification recheck policy (agent-loaded; do not invoke).
user-invocable	false

Verification Recheck Policy

Minimum Matrix

Style-only change: lint
Logic refactor (behavior-preserving intent): lint + functional smoke/regression + RTL-vs-RTL equivalence
Interface-impact change: lint + cdc + functional regression (+ equivalence when change is declared behavior-preserving)
Constraint/synthesis-impact change: lint + cdc + functional + synthesis/timing rerun + equivalence (RTL-vs-netlist or RTL-vs-RTL)

Recommended Commands (open-source baseline)

Lint:
- lint/scripts/run_lint.sh --tool verilator -f rtl/filelist_top.f --outdir lint/lint
CDC:
- lint/scripts/run_cdc.sh --tool structural --top <top> -f rtl/filelist_top.f --outdir lint/cdc
Functional smoke/regression:
- scripts/run_sim.sh --sim verilator --top <tb_top> -f rtl/filelist_top.f --outdir sim/reports
- bash {plugin_root}/skills/rtl-p5s-func-verify/scripts/run_regression.sh --mode local --seeds "1 42 123 1337 65536" --sim verilator ({plugin_root} = plugin root resolved from .rat/state/spawn-context.json)
Synthesis/timing recheck:
- syn/scripts/run_syn.sh --tool yosys --top <top> -f rtl/filelist_top.f
- If STA wrapper exists: syn/scripts/run_sta.sh --tool opensta --top <top> --outdir syn/rpt
Equivalence recheck:
- Delegate to equivalence-checker with context:
  - RTL-vs-RTL after refactor/ECO (reference=<pre-change>, implementation=<post-change>)
  - RTL-vs-netlist after synthesis-impact change (reference=rtl, implementation=syn/netlist.v)

Pass/Fail Criteria

lint: zero errors
cdc: no unwaived VIOLATION
functional: all must-pass scenarios green
regression: all required seeds pass (or documented waiver)
synthesis/timing: run completes with no fatal tool error
equivalence: no unresolved non-equivalent points (counterexamples must be resolved or approved as intentional deltas)

Escalation

Same category fails twice after fix attempts: escalate with replay script path
CDC or synthesis/timing failure after interface change: escalate to design owner
Missing replay artifacts: re-run command to generate reproducible evidence
Equivalence FAIL/UNKNOWN after behavior-preserving claim: escalate immediately (potential unintended semantic change)

Output Format

# Recheck Report
- Change Type: [style|logic|interface|constraint]
- Verdict: PASS | FAIL

## Executed Checks
| Check | Command | Result | Artifact |
|---|---|---|---|

## Failures and Escalation
- [if any]

المزيد من هذا المستودع

نفس المستودع

arch-review

babyworm/rtl-agent-team

Read-only architecture review of RTL vs uArch spec with area/timing/power tradeoffs. Use for post-RTL architecture sign-off or suspected spec mismatch.

2026-07-1742

bfm-develop

babyworm/rtl-agent-team

Builds Phase 3 SystemC TLM-2.0 bus functional models with AMBA (AXI/AHB/APB) support. Use for 'develop BFM', 'SystemC TLM model', 'bus functional model'.

2026-07-1742

codec-rd-eval

babyworm/rtl-agent-team

Encoder RD evaluation of C ref model: BD-PSNR/BD-rate (VCEG-M33) across sequences/QPs — 'RD eval', 'BD-rate', codec algorithm quality comparison.

2026-07-1742

cross-phase-contract-validator

babyworm/rtl-agent-team

Validate spec contracts across phase boundaries (P3 uarch to P4 RTL to P5 verify): port widths, memory class, REQ traceability. Use at P4/P5 entry.

2026-07-1742

p1-spec-research-policy

babyworm/rtl-agent-team

Internal reference: p1 spec research policy (agent-loaded; do not invoke).

2026-07-1742

rat-auto-design

babyworm/rtl-agent-team

Full 6-phase RTL pipeline from spec to verified design (P1 research through P6 design note). Triggers: 'design a chip', 'full pipeline', 'autopilot'.

2026-07-1742

Verification Recheck Policy

Minimum Matrix

Style-only change: lint

Logic refactor (behavior-preserving intent): lint + functional smoke/regression + RTL-vs-RTL equivalence

Interface-impact change: lint + cdc + functional regression (+ equivalence when change is declared behavior-preserving)

Constraint/synthesis-impact change: lint + cdc + functional + synthesis/timing rerun + equivalence (RTL-vs-netlist or RTL-vs-RTL)

Recommended Commands (open-source baseline)

Lint:

lint/scripts/run_lint.sh --tool verilator -f rtl/filelist_top.f --outdir lint/lint

CDC:

lint/scripts/run_cdc.sh --tool structural --top <top> -f rtl/filelist_top.f --outdir lint/cdc

Functional smoke/regression:

scripts/run_sim.sh --sim verilator --top <tb_top> -f rtl/filelist_top.f --outdir sim/reports
bash {plugin_root}/skills/rtl-p5s-func-verify/scripts/run_regression.sh --mode local --seeds "1 42 123 1337 65536" --sim verilator ({plugin_root} = plugin root resolved from .rat/state/spawn-context.json)

Synthesis/timing recheck:

syn/scripts/run_syn.sh --tool yosys --top <top> -f rtl/filelist_top.f
If STA wrapper exists: syn/scripts/run_sta.sh --tool opensta --top <top> --outdir syn/rpt

Equivalence recheck:

Delegate to equivalence-checker with context:
- RTL-vs-RTL after refactor/ECO (reference=<pre-change>, implementation=<post-change>)
- RTL-vs-netlist after synthesis-impact change (reference=rtl, implementation=syn/netlist.v)

Pass/Fail Criteria

lint: zero errors

cdc: no unwaived VIOLATION

functional: all must-pass scenarios green

regression: all required seeds pass (or documented waiver)

synthesis/timing: run completes with no fatal tool error

equivalence: no unresolved non-equivalent points (counterexamples must be resolved or approved as intentional deltas)

Escalation

Same category fails twice after fix attempts: escalate with replay script path

CDC or synthesis/timing failure after interface change: escalate to design owner

Missing replay artifacts: re-run command to generate reproducible evidence

Equivalence FAIL/UNKNOWN after behavior-preserving claim: escalate immediately (potential unintended semantic change)

Output Format

# Recheck Report - Change Type: [style|logic|interface|constraint] - Verdict: PASS | FAIL ## Executed Checks | Check | Command | Result | Artifact | |---|---|---|---| ## Failures and Escalation - [if any]