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/destynova2/cli-code-skills/cli-forge-perfnpx skills add Destynova2/cli-code-skills --skill cli-forge-perfgit clone --depth 1 https://github.com/Destynova2/cli-code-skillsWhat 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 | $0.00229 | $0.06707 |
| Opus 5 | $0.00114 | $0.03354 |
| Sonnet 5 | $0.00046 | $0.01341 |
| Haiku 4.5 | $0.00023 | $0.00671 |
Grade A, and why
cli-forge-perf 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 — 334 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Optimization: Ce skill utilise du chargement à la demande. Les contenus lourds vivent dans
references/et sont lus à la demande.
Language rule: Les instructions du skill sont en français (le catalogue lui-même). Pour les artefacts générés (rapports, benches, scripts), détecte la langue du projet (README, commentaires, docs, commits) et produis dans cette langue. Si le projet est bilingue, demande à l'utilisateur.
Gotchas: Lis
../gotchas.mdAVANT de produire un livrable.
Perf Optimization
Catalogue ordonné des techniques d'optimisation, du plus rentable au moins rentable. L'objectif : donner le bon réflexe au bon niveau, et empêcher de micro-optimiser avant d'avoir réglé l'algo et l'I/O.
Règle d'or : MESURE avant d'optimiser
"Premature optimization is the root of all evil." — Knuth
On n'optimise jamais à l'intuition. 90 % du temps est passé dans 10 % du code (le hot path). Optimiser le reste = effort gaspillé + complexité ajoutée pour rien. Méthode en boucle :
- Mesure — profile/trace pour trouver le vrai bottleneck (pas celui qu'on
imagine). Établis une baseline chiffrée. Ne suppose jamais quel langage/lib
est plus rapide : instrumente. Méthodo détaillée :
references/profiling.md. - Localise & classe — identifie LE point chaud ET sa nature : CPU-bound, memory-bound, bandwidth-bound, I/O-bound, lock-bound ou syscall-bound ? Cette classification (roofline) dicte quel niveau de la hiérarchie attaquer. Loi d'Amdahl : accélérer ×10 un bloc qui pèse 5 % du temps total ne gagne que ~4,5 %. Inutile.
- Optimise — applique la technique du bon niveau (voir hiérarchie ci-dessous).
- Re-mesure — confirme le gain réel. Si nul ou négatif, reviens en arrière. Garde le code lisible si le gain n'est pas significatif.
Reproductibilité — un bench n'a de valeur que s'il est rejouable à l'identique : seed, horloge, environnement, ordre, locale. Le toolkit canonique est partagé :
../shared/determinism.md. Sans ces pins, deux exécutions consécutives produisent des chiffres différents et l'A/B devient impossible à départager.
What ships with it
20 files 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.
- references/_project-profiles.md 3.7 KB
- references/algorithmic.md 6.1 KB
- references/async-concurrency.md 4.9 KB
- references/bench-protocol.md 7.3 KB
- references/benchmarking-traps.md 8.1 KB
- references/bio-inspired.md 5.0 KB
- references/cost-accounting.md 4.3 KB
- references/crypto-throughput.md 4.8 KB
- references/database.md 5.4 KB
- references/dependencies.md 4.3 KB
- references/experiment-method.md 14 KB
- references/frontend-web.md 5.9 KB
- references/idea-generation.md 4.5 KB
- references/lateral-thinking.md 7.1 KB
- references/llm-inference.md 8.1 KB
- references/math-physics.md 6.2 KB
- references/memory-cache.md 5.2 KB
- references/profiling.md 5.0 KB
- references/systems-hardware.md 6.6 KB
- references/worked-examples.md 5.7 KB
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 · 334 lines · 229 tokens per session scan A 296b13ba4485
cli-forge-perf is a skill published in the GitHub repository Destynova2/cli-code-skills (5 stars, last pushed 10d ago), licensed MIT. It adds 229 tokens to every session and 6,707 once invoked, about $0.0011 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 2026-08-31.
Other skills, from other repositories
clean-architecture
Apply Robert C. Martin's Clean Architecture, SOLID, and component-decoupling principles when designing, reviewing, or refactoring code — scaled correctly to the size of the task so trivial changes don't get over-engineered. Use this whenever the user asks to design a new feature/service, review code for architectural…
sdlc
Full SDLC workflow for implementing features, fixing bugs, refactoring code, testing, releasing, publishing, and deploying. Use this skill when implementing, fixing, refactoring, testing, adding features, building new code, or releasing/publishing/deploying.
feedback
Submit feedback, bug reports, feature requests, or share SDLC patterns you've discovered. Privacy-first — always asks before scanning.
deploy
Deploy the application to production.
claude-setup-wizard
Setup wizard — scans codebase, builds confidence per data point, only asks what it can't figure out, generates SDLC files. Use for first-time setup or re-running setup.
claude-update-wizard
Smart update for SDLC wizard — shows changelog, compares files, lets you selectively adopt changes while preserving customizations.