Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add skills/fl-sean03/opensdl/debug-runnpx skills add fl-sean03/OpenSDL --skill debug-rungit clone --depth 1 https://github.com/fl-sean03/OpenSDLWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00035 | $0.00378 |
| Opus 5 | $0.00017 | $0.00189 |
| Sonnet 5 | $0.00007 | $0.00076 |
| Haiku 4.5 | $0.00003 | $0.00038 |
Grade A, and why
debug-run scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
Debug a failed run
Inputs
- persisted run identifier
- manifest path
- workflow and inputs when a simulation reproduction is possible
Procedure
- Run
.agents/skills/debug-run/run.sh RUN_ID [MANIFEST]. - Run
uv run --locked opensdl events --manifest MANIFEST --run-id RUN_IDand identify the first failed task. - Trace the policy decision, lease, adapter call, timeout, retry, and error events.
- Reproduce with a simulation manifest when the workflow and inputs are available.
- Fix the smallest correct contract, adapter, runtime, storage, or configuration layer.
- Add a regression test and a conformance case when an adapter contract changed.
- Re-run the simulation and export it with
uv run --locked opensdl export RUN_ID --manifest MANIFEST. - Submit a repaired workflow as a new run with
opensdl run WORKFLOW --supersedes RUN_ID. A run records the workflow it was asked to execute, so--run-id RUN_IDresumes only the identical definition and is refused for any other; superseding leaves the failed run's record intact and names the replacement on both. - Check propagation impact before completion.
Completion
The cause is tied to persisted evidence, the regression test fails before the fix and passes after it, and the simulation rerun has an exported evidence bundle.
Stop conditions
Stop on ambiguous physical state or when safe recovery needs hold, abort, resume, or reconciliation behavior that the typed interface does not expose. Do not infer physical reversal from database state.
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 3d ago First seen · 40 lines · 35 tokens per session scan A 7a390cb75cdf
debug-run is a skill published in the GitHub repository fl-sean03/OpenSDL (102 stars, last pushed 3d ago), licensed Apache-2.0. It adds 35 tokens to every session and 378 once invoked, about $0.0002 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other skills, from other repositories
scientific-tool-onboarding
Use when evaluating, adding, or upgrading ScholarAIO support for a scientific computing tool, especially integration-gate review, official docs ingestion, toolref integration, lightweight skill design, and end-to-end CLI verification.
bioinformatics
Use when working on bioinformatics workflows such as alignment, variant calling, phylogenetics, or protein-structure analysis, especially across BLAST, minimap2, samtools, bcftools, MAFFT, IQ-TREE, or ESMFold.
lammps
Use when working on classical materials simulations with LAMMPS, including potential selection, shock or deformation setups, thermodynamic runs, or structure analysis for solids and nanomaterials.
paper-guided-reading
Use when the user wants guided reading, deep reading, or structured analysis of a single paper, starting from fuzzy keywords or a research question.
paper-writing
Use when the user wants help drafting or revising a research paper section, including Introduction, Related Work, Method, Results, Discussion, or Conclusion.
paper2any
Use when the user explicitly requests Paper2Any or a Paper2Any benchmark for paper-to-figure, PPT, poster, video, DrawIO, PDF-to-PPT, or related workflows through the isolated ScholarAIO sidecar.