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/sordi-ai/skill-everything/debuggingnpx skills add sordi-ai/skill-everything --skill debugginggit clone --depth 1 https://github.com/sordi-ai/skill-everythingWhat 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 | $0.00817 |
| Opus 5 | $0.00017 | $0.00409 |
| Sonnet 5 | $0.00007 | $0.00163 |
| Haiku 4.5 | $0.00003 | $0.00082 |
Grade A, and why
debugging 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 2d 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.
How it starts
The opening of the file, as written. The whole thing — 61 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Sub-Skill: Debugging
Purpose: Systematic techniques for reproducing failures, isolating variables, and finding root causes without guessing. Complements skills/tdd/SKILL.md — a failing test is the best reproduction case.
Rules
Reproduction
- Reproduce before fixing. Before changing any code, confirm you can trigger the failure on demand. A fix applied to an unreproduced bug is a guess, not a solution. Reference: ERR-2026-018
- Capture the minimal reproduction. Reduce the failing scenario to the smallest input, fewest dependencies, and shortest code path that still exhibits the symptom. Smaller reproductions expose the cause faster and prevent regression.
- Write the reproduction as a test. Encode the reproduction case as an automated test before touching production code. This locks in the failure signal and prevents silent regression. See
skills/tdd/SKILL.md.
Isolation
- Change one variable at a time. Never modify multiple suspects simultaneously. Each change must be independently observable so you know which variable caused the change in behaviour.
- Use binary search on the call stack. When the failure source is unknown, bisect: confirm the bug is present at the midpoint of the execution path, then recurse into the half that contains it. Avoid reading every line top-to-bottom.
- Prefer structured logging over print-debugging. Add log statements at decision boundaries with structured key-value pairs (not free-form strings). Structured output is grep-able, diffable, and removable without side effects.
Hypothesis and Root Cause
- State a falsifiable hypothesis before each experiment. Write down: "I believe X causes Y because Z. If I change X, Y should disappear." Run the experiment. If the hypothesis is wrong, update your model before the next experiment.
- Distinguish symptom from cause. The error message or stack trace is the symptom. The root cause is the incorrect assumption, missing guard, or wrong state that produced it. Never stop at the symptom — trace back to the decision that allowed the bad state to exist.
- Do not fix symptoms in isolation. Patching the symptom without addressing the root cause produces a second bug that hides the first. Ensure the fix makes the root cause impossible, not just the observed symptom unlikely.
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.
- 2d ago First seen · 61 lines · 35 tokens per session scan A db805c516a3c
debugging is a skill published in the GitHub repository sordi-ai/skill-everything (19 stars, last pushed 3mo ago), licensed MIT. It adds 35 tokens to every session and 817 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
validate-context-routes
Validate that literal project-relative paths referenced by a project's root AGENTS.md exist. Use when changing AGENTS.md routing, moving shared documentation or .agents resources, or checking a dotagents example.
save-context
Use this skill at milestone handoff, session close, or when the user asks Atlas to preserve state.
chroma-hybrid-search
Local hybrid retrieval (BM25 + ChromaDB vector + BGE-Reranker) over /.deep-memory hot and cold stores. Use when high-accuracy code/solution retrieval is needed and hallucination must be minimized. Typically invoked by deep-memory.
cline-fix-volatile-msg
Ladder-aware Cline Anthropic caching — verify the rolling read/write ladder on the wire, then add the tools breakpoint and tune TTL. Updated for the 2026-08 AI-SDK monorepo.
continue-gemini-explicit
Continue's Gemini provider doesn't use the cachedContents API at all. Add explicit caching for sessions over the minimum token threshold.
aider-1h-ttl
Aider uses 5min TTL by default and works around long pauses with keepalive pings. Wire up the 1h TTL beta instead.