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 skills add OutlineDriven/odin-gemini-cli-extension --skill perf-profilegit clone --depth 1 https://github.com/OutlineDriven/odin-gemini-cli-extensionWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/outlinedriven/odin-gemini-cli-extension/perf-profile)<a href="https://agentmods.dev/skills/outlinedriven/odin-gemini-cli-extension/perf-profile"><img src="https://agentmods.dev/badge/skills/outlinedriven/odin-gemini-cli-extension/perf-profile/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/outlinedriven/odin-gemini-cli-extension/perf-profile"><img src="https://agentmods.dev/badge/skills/outlinedriven/odin-gemini-cli-extension/perf-profile.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What 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 5.1 | $0.00000 | $0.01077 |
| Opus 5 | $0.00000 | $0.00539 |
| Sonnet 5 | $0.00000 | $0.00215 |
| Haiku 4.5 | $0.00000 | $0.00108 |
Grade A, and why
perf-profile 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 9d 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.
This is a copy
100% identical to perf-profile — 0 lines differ, which has more behind it and is treated as the original. This page carries a canonical link to it rather than competing with it.
How it starts
The opening of the file, as written. The whole thing — 66 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Performance is a contract with reality. Intuition about hot paths is wrong more often than right. Capture, locate, hypothesize, optimize, re-measure, prove the regression with a benchmark — then defend the win with an invariant.
When to Apply / NOT
Apply: latency or throughput SLO violation; memory pressure (RSS growth, GC churn); benchmark regression; pre-optimization scoping; cold-start vs steady-state cost split; cache locality / branch-prediction concerns.
NOT apply: defect with wrong outputs; architectural redesign; micro-optimization without budget pressure; untested code (write tests first).
Anti-patterns
- Optimize without profile: intuition-driven changes.
- Single-run benchmarks: variance dominates. Use
hyperfine --warmup 3 --min-runs 10. - Profile in debug build: optimizer-disabled binaries lie.
- Confuse flat profile with call-graph: self-time vs total-time tell different stories.
- Ignoring tail latency: p50 stays flat while p99 explodes.
- Cherry-picking the win: re-measure end-to-end.
- Allocation blind spot: CPU profiler hides GC.
- Forgetting the regression guard.
Workflow (language-neutral)
- Define budget — restate target metric: latency p95 < X ms, throughput > Y rps, RSS < Z MB.
- Establish baseline — run unoptimized workload under
hyperfineplus profiler. Save raw artifacts. - Capture profile — sampled CPU profile → flamegraph; allocation profile if memory-bound; latency histogram for tail.
- Locate hotspot — top self-time function or widest plateau. Cross-check with allocation profile.
- Hypothesize — one falsifiable claim with predicted delta.
- Optimize minimally — smallest change targeting the hypothesis.
- Re-profile — capture same metric; differential flamegraph.
- Prove the win —
hyperfine 'baseline' 'optimized' --warmup 3 --min-runs 10. - Guard the win — add CI benchmark with regression bound.
Reading Flamegraphs
- Wide plateau on top: hot self-time function — primary target.
- Narrow towers: deep call chains — examine for over-abstraction.
- Repeated motifs: same callee under many parents — candidate for inlining or caching.
- Missing frames: rebuild with
-fno-omit-frame-pointer/RUSTFLAGS=-C force-frame-pointers=yes. - Differential flamegraph (
hotspot --diff): red = added cost, blue = removed.
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.
- 9d ago First seen · 66 lines · 0 tokens per session scan A c3b66bcdd461
perf-profile is a skill published in the GitHub repository OutlineDriven/odin-gemini-cli-extension (5 stars, last pushed 2mo ago), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 1,077 tokens. A static security scan graded it A with 0 findings. It is 100% identical to perf-profile, differing in 0 lines, and is treated as a copy.
Other skills, from other repositories
design
Set visual and interaction direction for any UI surface (web, React, TUI, CLI, desktop, Qt, design-system tokens) before any UI code. Direction-first: generates 3-4 distinct directions via verbalized sampling, picks one via per-axis single-select, then derives palette, typography, spacing, motion budget. Loads when…
askme
Verbalized Sampling (VS) protocol for intent exploration before planning, mode-aware. Default exhaustive runs full VS; collaborative runs tip-sharing dialogue; adversarial walks the design tree one fork at a time. Auto-detects from phrasing ("help me refine" → collaborative, "poke holes" → adversarial); override via…
git-branchless
Enforce idiomatic git-branchless during planning and executing tasks — detached-HEAD-first work, in-memory rebase via git move, event-log recovery via git undo, deferred branch creation, speculative-merge git sync for base updates. Use when planning or executing multi-commit work, history rewrites, stack edits…
grill-ai-mastery
Hybrid interview that probes AI-engineering mastery by tip-vocabulary depth — entity referencing, loop closure, observability, harness improvement — not by token usage or LOC. Start collaborative (two-way tip exchange), escalate to adversarial probing when depth is lacking. Trigger when the user says "interview me on…
grill-me
Adversarial relentless interview against any plan or design until shared understanding is reached. Walk the decision tree, resolve dependencies one answer at a time, recommend a default per question. Trigger when the user says "grill me", "stress-test this", "interview me about this design", or otherwise asks for…
proof-driven
Proof-driven development. Use when implementing with formal verification using property-based testing, theorem proving, or proof tactics; zero unproven property policy enforced.