hipDNN Deep Review
$SKILL is the directory this skill was loaded from (the directory containing this
SKILL.md). You are the orchestrator: execute RUNBOOK.md. You pin the change, dispatch
scouts, lanes and verifiers, merge their output and write the report; you do not review
code. Your formats are in report.md; RUNBOOK.md Fan out says
what each agent reads, and lanes/ holds the briefs. RUN builds with
hipdnn-superbuild and tests with
hipdnn-superbuild-test; IMPACT runs the scripts of
hipdnn-ingestor-impact. Install those skills alongside this one.
Prerequisites: network access, an authenticated gh with read access, git 2.38 or later (pin.py uses git merge-tree --write-tree), and Python 3 (it runs the scripts under
$SKILL/scripts/); pre-commit and pytest for the lanes that run validators. The IMPACT lane needs
Python 3.10 or newer and PyYAML; its emitter also needs Jinja2 (hipdnn-ingestor-impact Prerequisites). RUN and verifiers also need
a ROCm install (Linux: /opt/rocm or ROCM_PATH; Windows: the wheel SDK that hipdnn-superbuild provisions), CMake, Ninja and about 25 GB free in $SCRATCH; without them defect and
test-gap findings stay UNVERIFIED and the review closes only through human
ACCEPTED-RISK decisions, recorded with ledger.py set (PIN.md records Build host).
Platforms: Linux, macOS and Windows. On Windows run the skill from Git Bash (Git for Windows): the runbook's shell blocks need
bash, awk, sed, grep and comm. Blocks write the interpreter as python3; on Windows that name is the Microsoft Store
stub, so use python or py -3 instead. Scripts read and write UTF-8 and LF whatever the host's locale.
For a quick answer without a review, use hipdnn-ingestor-impact: it runs the
same scripts as the IMPACT lane (lanes/runtime-impact.md) and ends in a Runtime impact: line.
Entry contract
Ask for what is missing; record it in PIN.md:
| Input |
Notes |
| Target |
PR URL or number, branch:<name>, or <base>..<head>; intent:<file> (what it is for, how tested, Known issues; required without a PR) holds the intent document, whose Known issues section lists open items the author already knows, numbered K1, K2 and so on, and never renumbered; without a PR, M4 and M5 are N/A (no PR) (policy.md M-rules) |
evidence:<dir>, scratch:<dir>, clone:<dir> |
$EVIDENCE outside the product tree; disposable $SCRATCH; optional clone (default $SCRATCH/clone) |
terms:<file>, knowledge:<dir> |
Optional private terms, one fixed string per line, outside the repository; optional $KNOWLEDGE store |
round:<N> |
Default: one more than the highest $EVIDENCE/r<k>/; 1 when none |
prior:<dir> |
Optional published record of earlier rounds that $EVIDENCE lacks (LEDGER.md and r<k>/); RUNBOOK.md Iterate seeds $EVIDENCE from it |
| Agent limits |
Ask the user or the host for Agents: (maximum concurrent agents), Cores: (CPU cores for builds and tests) and Mutation: (a host recipe for mutating a built tree, or none); the skill assumes no values. They go into dispatch-context.md (RUNBOOK.md Fan out) |
| Scope notes |
Ask the user for scope notes the change's description does not state, such as areas being deprecated, or none. Free text; it goes into dispatch-context.md as Scope: and is informational unless it says "do not file findings in " (lanes/common.md Rules) |
| Hazard knowledge |
Optional: a directory of engine-applicability hazard notes (distinct from knowledge:), or none. It goes into dispatch-context.md as Hazard knowledge: and is read by the APPL lane (lanes/applicability.md Hazard knowledge directory) |
Local builds and host-only tests are allowed. GPU, remote or shared compute, and every write
to GitHub, Jira or a remote branch need the user's approval for that exact action. The
review is advisory: the user posts the draft comments. Write only $EVIDENCE, $KNOWLEDGE
and $SCRATCH, never the product tree.
Completion
A round is complete when its gates hold and REPORT.md and LEDGER.md exist for the head.
The review is closed (Closure: CLOSED) when no BLOCKING or IMPORTANT finding is OPEN
or UNVERIFIED and no STOP or decision is unresolved. Otherwise Closure: OPEN lists what
blocks it and Iterate starts the next round; the user decides when to stop.
1---2name: hipdnn-deep-review3description: "Multi-agent deep review of a hipDNN or dnn-providers PR, branch or commit range. Scouts map each affected area, focused reviewer lanes run in parallel, every finding is verified by source proof, reproduction or mutation, and a ledger carries findings across rounds until no blocking or important one stays open. Its IMPACT lane reports kernel ingestor changes that break at run time while the build stays green (kernel ABI, dispatch, descriptors) with the scripts of hipdnn-ingestor-impact, which answers that question without a review. For a PR body or a quick single-pass review, use hipdnn-pr-quality."4argument-hint: "[<PR URL or number> | branch:<name> | <base>..<head>] evidence:<dir> scratch:<dir> [clone:<dir>] [intent:<file>] [terms:<file>] [knowledge:<dir>] [round:<N>] [prior:<dir>]"5allowed-tools: Bash, Read, Write, Edit, Grep, Glob, Task6metadata:7 source: https://skillmd.com/skills/rocm/hipdnn-deep-review8---910# hipDNN Deep Review1112`$SKILL` is the directory this skill was loaded from (the directory containing this13`SKILL.md`). You are the orchestrator: execute [RUNBOOK.md](RUNBOOK.md). You pin the change, dispatch14scouts, lanes and verifiers, merge their output and write the report; you do not review15code. Your formats are in [report.md](report.md); [RUNBOOK.md](RUNBOOK.md) *Fan out* says16what each agent reads, and `lanes/` holds the briefs. `RUN` builds with17[hipdnn-superbuild](../hipdnn-superbuild/SKILL.md) and tests with18[hipdnn-superbuild-test](../hipdnn-superbuild-test/SKILL.md); `IMPACT` runs the scripts of19[hipdnn-ingestor-impact](../hipdnn-ingestor-impact/SKILL.md). Install those skills alongside this one.2021Prerequisites: network access, an authenticated `gh` with read access, `git` 2.38 or later (`pin.py` uses `git merge-tree --write-tree`), and Python 3 (it runs the scripts under22`$SKILL/scripts/`); `pre-commit` and `pytest` for the lanes that run validators. The `IMPACT` lane needs23Python 3.10 or newer and PyYAML; its emitter also needs Jinja2 ([hipdnn-ingestor-impact](../hipdnn-ingestor-impact/SKILL.md) *Prerequisites*). `RUN` and verifiers also need24a ROCm install (Linux: `/opt/rocm` or `ROCM_PATH`; Windows: the wheel SDK that [hipdnn-superbuild](../hipdnn-superbuild/SKILL.md) provisions), CMake, Ninja and about 25 GB free in `$SCRATCH`; without them `defect` and25`test-gap` findings stay `UNVERIFIED` and the review closes only through human26`ACCEPTED-RISK` decisions, recorded with `ledger.py set` (`PIN.md` records `Build host`).2728Platforms: Linux, macOS and Windows. On Windows run the skill from Git Bash (Git for Windows): the runbook's shell blocks need29`bash`, `awk`, `sed`, `grep` and `comm`. Blocks write the interpreter as `python3`; on Windows that name is the Microsoft Store30stub, so use `python` or `py -3` instead. Scripts read and write UTF-8 and LF whatever the host's locale.3132For a quick answer without a review, use [hipdnn-ingestor-impact](../hipdnn-ingestor-impact/SKILL.md): it runs the33same scripts as the `IMPACT` lane ([lanes/runtime-impact.md](lanes/runtime-impact.md)) and ends in a `Runtime impact:` line.3435## Entry contract3637Ask for what is missing; record it in `PIN.md`:3839| Input | Notes |40|---|---|41| Target | PR URL or number, `branch:<name>`, or `<base>..<head>`; `intent:<file>` (what it is for, how tested, `Known issues`; required without a PR) holds the intent document, whose `Known issues` section lists open items the author already knows, numbered `K1`, `K2` and so on, and never renumbered; without a PR, M4 and M5 are `N/A (no PR)` ([policy.md](policy.md) *M-rules*) |42| `evidence:<dir>`, `scratch:<dir>`, `clone:<dir>` | `$EVIDENCE` outside the product tree; disposable `$SCRATCH`; optional clone (default `$SCRATCH/clone`) |43| `terms:<file>`, `knowledge:<dir>` | Optional private terms, one fixed string per line, outside the repository; optional `$KNOWLEDGE` store |44| `round:<N>` | Default: one more than the highest `$EVIDENCE/r<k>/`; 1 when none |45| `prior:<dir>` | Optional published record of earlier rounds that `$EVIDENCE` lacks (`LEDGER.md` and `r<k>/`); [RUNBOOK.md](RUNBOOK.md) *Iterate* seeds `$EVIDENCE` from it |46| Agent limits | Ask the user or the host for `Agents:` (maximum concurrent agents), `Cores:` (CPU cores for builds and tests) and `Mutation:` (a host recipe for mutating a built tree, or none); the skill assumes no values. They go into `dispatch-context.md` ([RUNBOOK.md](RUNBOOK.md) *Fan out*) |47| Scope notes | Ask the user for scope notes the change's description does not state, such as areas being deprecated, or `none`. Free text; it goes into `dispatch-context.md` as `Scope:` and is informational unless it says "do not file findings in <area>" ([lanes/common.md](lanes/common.md) *Rules*) |48| Hazard knowledge | Optional: a directory of engine-applicability hazard notes (distinct from `knowledge:`), or none. It goes into `dispatch-context.md` as `Hazard knowledge:` and is read by the `APPL` lane ([lanes/applicability.md](lanes/applicability.md) *Hazard knowledge directory*) |4950Local builds and host-only tests are allowed. GPU, remote or shared compute, and every write51to GitHub, Jira or a remote branch need the user's approval for that exact action. The52review is advisory: the user posts the draft comments. Write only `$EVIDENCE`, `$KNOWLEDGE`53and `$SCRATCH`, never the product tree.5455## Completion5657A round is complete when its gates hold and `REPORT.md` and `LEDGER.md` exist for the head.58The review is closed (`Closure: CLOSED`) when no `BLOCKING` or `IMPORTANT` finding is `OPEN`59or `UNVERIFIED` and no `STOP` or decision is unresolved. Otherwise `Closure: OPEN` lists what60blocks it and *Iterate* starts the next round; the user decides when to stop.