Swig Test
swig/swig
Run SWIG test suite for specific languages. An agent skill from swig/swig.
Describes a decision tree of steps and skills to utilize when, starting from a test failure or the failure of bazel run command using heir-opt, you would like to produce a reproducing input IR that…
Install Reproduce Failure
$ npx skills add google/heir --skill reproduce-failure -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install google/heir reproduce-failure --agent claude-codeProject scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).
$ git clone --depth 1 https://github.com/google/heir.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.agents/skills/reproduce_failure .claude/skills/reproduce-failure && rm -rf skills-srcUse ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.
Claude Code skills documentation · loads skills from .claude/skills/
Install the "reproduce-failure" agent skill from https://github.com/google/heir/tree/main/.agents/skills/reproduce_failure into .claude/skills/reproduce-failure/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "reproduce-failure", then confirm the skill loads.Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$skill-installer install https://github.com/google/heir/tree/main/.agents/skills/reproduce_failureType this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.
$ npx skills add google/heir --skill reproduce-failure -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install google/heir reproduce-failure --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/google/heir.git skills-src && mkdir -p .agents/skills && cp -r skills-src/.agents/skills/reproduce_failure .agents/skills/reproduce-failure && rm -rf skills-srcUse ~/.agents/skills/ instead of .agents/skills for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "reproduce-failure" agent skill from https://github.com/google/heir/tree/main/.agents/skills/reproduce_failure into .agents/skills/reproduce-failure/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "reproduce-failure", then confirm the skill loads.Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add google/heir --skill reproduce-failure -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install google/heir reproduce-failure --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/google/heir.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/.agents/skills/reproduce_failure .cursor/skills/reproduce-failure && rm -rf skills-srcUse ~/.cursor/skills/ instead of .cursor/skills for a personal install.
Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/
Install the "reproduce-failure" agent skill from https://github.com/google/heir/tree/main/.agents/skills/reproduce_failure into .cursor/skills/reproduce-failure/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "reproduce-failure", then confirm the skill loads.Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gemini skills install https://github.com/google/heir.git --path .agents/skills/reproduce_failure--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.
$ npx skills add google/heir --skill reproduce-failure -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install google/heir reproduce-failure --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/google/heir.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/.agents/skills/reproduce_failure .gemini/skills/reproduce-failure && rm -rf skills-srcUse ~/.gemini/skills/ instead of .gemini/skills for a personal install, then run /skills reload.
Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/
Install the "reproduce-failure" agent skill from https://github.com/google/heir/tree/main/.agents/skills/reproduce_failure into .gemini/skills/reproduce-failure/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "reproduce-failure", then confirm the skill loads.Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gh skill install google/heir reproduce-failureInstalls for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).
$ npx skills add google/heir --skill reproduce-failure -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
$ git clone --depth 1 https://github.com/google/heir.git skills-src && mkdir -p .github/skills && cp -r skills-src/.agents/skills/reproduce_failure .github/skills/reproduce-failure && rm -rf skills-srcUse ~/.copilot/skills/ instead of .github/skills for a personal install. Commit .github/skills so cloud agent and code review can use it.
GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/
Install the "reproduce-failure" agent skill from https://github.com/google/heir/tree/main/.agents/skills/reproduce_failure into .github/skills/reproduce-failure/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "reproduce-failure", then confirm the skill loads.GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add google/heir --skill reproduce-failure -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install google/heir reproduce-failure --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/google/heir.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/.agents/skills/reproduce_failure .opencode/skills/reproduce-failure && rm -rf skills-srcUse ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.
OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/
Install the "reproduce-failure" agent skill from https://github.com/google/heir/tree/main/.agents/skills/reproduce_failure into .opencode/skills/reproduce-failure/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "reproduce-failure", then confirm the skill loads.OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
Facts
- Skill name
reproduce-failure- GitHub stars
- 929
- Token cost
- ~1.4k tokens
- SKILL.md length
- 634 words
- Files
- 1
- Skills in repo
- 4
- Repo updated
- First seen
- Licence
- Apache-2.0
At a glance
Describes a decision tree of steps and skills to utilize when, starting from a test failure or the failure of bazel run command using heir-opt, you would like to produce a reproducing input IR that…
- Works in 5 steps: A direct bazel run command. → A lit test target or bazel test command… → A file path to a lit test, such as → …
- Tasks that involve Failing and flaky tests
- SKILL.md covers Overview, Usage and Gotchas
- Calls bazel
What it does
Reproduce Failure is an agent skill from google/heir, published by the product's own GitHub organization. Describes a decision tree of steps and skills to utilize when, starting from a test failure or the failure of bazel run command using heir-opt, you would like to produce a reproducing input IR that fails on a particular compiler pass.
Its SKILL.md is about 1.4k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.
It sits in Testing & QA, covering Failing and flaky tests. The repository describes itself as: The compiler for homomorphic encryption. The licence is Apache-2.0.
When your agent uses it
- Tasks that involve Failing and flaky tests
Example prompts
- “Use the reproduce-failure skill to describe a decision tree of steps and skills to utilize when, starting from a test failure or the failure of…”
- “/reproduce-failure”
Workflow steps
5 steps, taken from the first numbered list in SKILL.md.
- A direct bazel run command.
- A lit test target or bazel test command pointing to such a target, such
- A file path to a lit test, such as
- An end-to-end test target, such as
- A filepath to an end-to-end test file, such as
What it can do on your machine
Read from SKILL.md and the folder at commit 26da825. It shows what the files ask for, not the result of running them.
Tool permissions
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
Runs code
Shell commands in SKILL.md call:
bazel
From the folder's file list and the shell code blocks in SKILL.md.
Network
No URLs in SKILL.md.
From URLs in SKILL.md, links to its own repository left out.
Credentials
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.
Context cost
Reproduce Failure loads about 1.4k tokens when it runs. Until then it costs about 64 tokens; SKILL.md has 634 words of instructions outside code blocks.
Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.
Safety
Auto-check passed
The automated check found no risky patterns in SKILL.md.
Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.
SKILL.md
The full file from google/heir at commit 26da825, republished under its Apache-2.0 licence (© google). 634 words, ~1,397 tokens.
Download SKILL.mdSave it as
.claude/skills/reproduce-failure/SKILL.md (or your agent's skills folder).- name
- reproduce-failure
- description
- Describes a decision tree of steps and skills to utilize when, starting from a test failure or the failure of `bazel run` command using heir-opt, you would like to produce a reproducing input IR that fails on a particular compiler pass.
Reproduce a test failure
Overview
This skill guides the agent in identifying an IR and compiler pass that
reproduces a failure from a lit or e2e test target, or from a bazel run
command that runs heir-opt.
Usage
To reproduce a failure, follow these steps.
Identify the starting point of the process
The user may input one of multiple "starting points" for a failure.
- A direct
bazel runcommand. - A
littest target orbazel testcommand pointing to such a target, such as//tests/Transforms/convert_to_ciphertext_semantics:assign_layout.mlir.test, - A file path to a lit test, such as
third_party/heir/tests/Transforms/convert_to_ciphertext_semantics/assign_layout.mlir - An end-to-end test target, such as
//tests/Examples/lattigo/ckks/dot_product_8f:dotproduct8f_test, - A filepath to an end-to-end test file, such as
third_party/heir/tests/Examples/lattigo/ckks/dot_product_8f/dot_product_8f_test.go
Reiterate to the user that you understand which of these options are the
starting point, and if the user provides a starting point that doesn't fit into
one of these options, warn them and then ask for confirmation when improvising
how to convert their failure to a bazel run command in the next step.
Identify the right bazel run command
For option 1 (A direct bazel run command), nothing is needed, continue to the
next step.
For option 2, use the lit_to_bazel skill.
For option 3, use the e2e_to_bazel skill.
At the end of this step, you should have a bazel run command that reproduces
the user's error, but that command may run many compiler passes in a pipeline.
The remaining steps will reduce the bazel run command to a more minimal
reproducer.
Run it to ensure it reproduces the user's error.
Identify the specific IR, pass, and pass options
Use the dump_intermediate_ir skill to augment the bazel run command with
appropriate flags, so that it writes relevant files that contain the IR and pass
options.
Then inspect the right file (usually the last file dumped before an error occurs) to identify the IR and pass options. The dumped IR should include near the top of the file, a comment like
// -----// IR Dump Before CanonicalizerPass: canonicalize{cse-between-iterations=false max-iterations=5 max-num-rewrites=-1 region-simplify=normal test-convergence=false top-down=true} //----- //The part after : is the relevant pass name and options that will be needed in
the next step. The rest of the file content is the IR that will be needed in the
next step.
Show full SKILL.md (273 more words)Show less
Construct the bazel run command for the individual pass
Given the information from previous steps, construct a bazel run command of
the form:
bazel run //tools:heir-opt -- \
--pass-pipeline="<PASS_AND_OPTIONS>" \
<IR_FILE>where <PASS_AND_OPTIONS> corresponds to the "pass name and options" from the
previous step, and <IR_FILE> corresponds to the dumped IR for that pass.
At this step, run the new bazel run command and make sure that it reproduces
the error.
Create a reproducing lit test file
Copy the IR to a new lit regression test file with the bazel run command
modified to a // RUN: line.
The new test should live in third_party/heir/tests/Regression/ and have at the
top of the file a comment describing the origin of the reproducer along with the
// RUN: line.
For example, if the reduced bazel run command from the previous step was
bazel run //tools:heir-opt -- \
--pass-pipeline="canonicalze{cse-between-iterations=true}" \
/tmp/mlir/foo.mlirand the original command was
bazel run //tools:heir-opt -- --big-pipeline /path/to/file.mlir
Then the corresponding lit file should start with
// This file is a minimal reproducer of a failure originally produced with
//
// heir-opt --big-pipeline /path/to/file.mlir
//
// RUN: heir-opt --pass-pipeline="canonicalze{cse-between-iterations=true}" %sBelow that header, place the reproducing IR from the previous step. Then confirm
the test exercises the test failure once more by running
bazel test //tests/Regression:<name_of_file.mlir>.test.
The <name_of_file.mlir> should be chosen appropriately according to the
following heuristic:
- If you know of a particular GitHub issue number, use the number, like
issue_1480.mlir. - If you have a starting test file or target, name it similar to that file,
e.g.,
dot_product_8f_regression.mlir - If it makes sense, incorporate the name of the pass, e.g.,
dot_product_secret_to_ckks.mlir - If no natural name makes sense, use
RENAME_ME_regression_test.mlir
Gotchas
- Ensure that at each step you can still reproduce the failure. If at any step, a newly reduced command fails to reproduce the failure, stop and report this to the user, asking for assistance.
Copy this checklist and track progress:
- [ ] Step 1: Identify the starting point of the process
- [ ] Step 2: Identify the right `bazel run` command
- [ ] Step 3: Identify the specific IR, pass, and pass options
- [ ] Step 4: Construct the `bazel run` command for the individual pass
- [ ] Step 5: Create a reproducing `lit` test file
Each step corresponds to a section mentioned above.<!-- mdformat global-off -->© google, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file
Files
Just SKILL.md in .agents/skills/reproduce_failure of google/heir.
Open the folder on GitHubat commit 26da825
Compare with similar skills
Reproduce Failure next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.
| Skill | Stars | Used in | Tokens | Auto-check | Licence | Repo updated |
|---|---|---|---|---|---|---|
| Reproduce Failure this skillgoogle/heir | 929 | — | ~1.4k | Automated safety check: Pass | Apache-2.0 | |
| Swig Testswig/swig | 6.3k | — | ~2.3k | Automated safety check: Pass | Custom licence | |
| Triage CI FailureDataDog/datadog-agent | 3.8k | — | ~2.3k | Automated safety check: Pass | Apache-2.0 | |
| Dynamo Jira TicketDynamoDS/Dynamo | 2k | — | ~1.1k | Automated safety check: Pass | Apache-2.0 | |
| Fix Ready PRsfastrepl/anarlog | 9.5k | — | ~1.4k | Automated safety check: Pass | MIT | |
| Trx Analysismicrosoft/vstest | 969 | — | ~1.8k | Automated safety check: Pass | MIT |
Similar skills
- Official
Triage CI Failure
DataDog/datadog-agent
Classify a failed CI as either caused by an active incident, flakiness, or a true code regression.
Dynamo Jira Ticket
DynamoDS/Dynamo
Create structured Jira tickets for Dynamo from bug reports, failing tests, or feature requests.
Fix Ready PRs
fastrepl/anarlog
Inspect every open non-draft PR for CI failures and unresolved Cursor Bugbot findings, then fix them on the existing PR branches.
- Official
Trx Analysis
microsoft/vstest
Parse and analyze Visual Studio TRX test result files. An agent skill from microsoft/vstest.
Wio
workersio/skills
Testing workflow skill for finding high-value test candidates, writing focused tests, generating realistic workloads, reviewing test value, and diagnosing test-suite health.
More from google/heir
- Official
Lit To Bazel
google/heir
Converts MLIR lit test files to bazel run commands. An agent skill from google/heir.
- Official
E2E To Bazel
google/heir
Converts end-to-end (e2e) test targets or paths to bazel run commands for heir-opt.
- Official
Dump Intermediate Ir
google/heir
Augments a bazel run CLI command for heir-opt with additional flags that dump the IR before each compiler pass, allowing the agent to inspect intermediate IR during compilation, identify a…
Related
Categories
Questions about Reproduce Failure
What does Reproduce Failure do?
Describes a decision tree of steps and skills to utilize when, starting from a test failure or the failure of bazel run command using heir-opt, you would like to produce a reproducing input IR that…. Reproduce Failure is an agent skill from google/heir, published by the product's own GitHub organization. Describes a decision tree of steps and skills to utilize when, starting from a test failure or the failure of bazel run command using heir-opt, you would like to produce a reproducing input IR that fails on a particular compiler pass.
When should I use Reproduce Failure?
Reproduce Failure fits situations like: tasks that involve Failing and flaky tests.
How do I install Reproduce Failure in Claude Code?
Run `npx skills add google/heir --skill reproduce-failure -a claude-code`. Or copy the skill folder (.agents/skills/reproduce_failure in google/heir) into .claude/skills/reproduce-failure in your project. Claude Code loads it when a task matches its description.
How do I install Reproduce Failure in Codex?
Run `npx skills add google/heir --skill reproduce-failure -a codex`. Or copy the skill folder (.agents/skills/reproduce_failure in google/heir) into .agents/skills/reproduce-failure in your project. Codex loads it when a task matches its description.
Can I use Reproduce Failure in Cursor, Gemini CLI or GitHub Copilot?
Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add google/heir --skill reproduce-failure -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/reproduce-failure, .gemini/skills/reproduce-failure, .github/skills/reproduce-failure and .opencode/skills/reproduce-failure in your project.
What does Reproduce Failure need to run?
Going by SKILL.md and its folder, Reproduce Failure needs the command-line tools its instructions call (bazel).
Does Reproduce Failure access the network?
SKILL.md contains no URLs. Any network use would come from the scripts or tools the agent runs. This is read from the text; nothing was executed.
Is Reproduce Failure safe to install?
Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.
What licence does Reproduce Failure use?
Reproduce Failure is published under the Apache-2.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.
How many tokens does Reproduce Failure use?
About 1.4k tokens (SKILL.md is roughly 5.6k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.
What are the alternatives to Reproduce Failure?
Skills that share tags, products or a category with Reproduce Failure: Swig Test (swig/swig, 6.3k stars), Triage CI Failure (DataDog/datadog-agent, 3.8k stars), Dynamo Jira Ticket (DynamoDS/Dynamo, 2k stars) and Fix Ready PRs (fastrepl/anarlog, 9.5k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
Who maintains Reproduce Failure?
google (a GitHub organization, an official publisher) maintains it in google/heir, which has 929 GitHub stars. The repository holds 4 skills in this directory. The repository was last updated on October 9, 2026.
Source: google/heir on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.