agentleFS
Sign inStart free
matthewdonsemail-labmatthewdonsemail-lab/log/skills/callTiming/SKILL.md

callTiming

Skill63 starsChanged yesterdayMITView on GitHub

Find where a log call spends its time from stored timings, never a guess (rule 165). Every MCP tool call and terminal run has a call id and stores each stage as a span in log.db; `timing` reports p50, p95 and p99 per tool and per stage, and `timing` { call } shows one call as a waterfall. Use when the owner asks why something is slow, whether we hit p99 or p95, or before and after any change made to speed something up.

The file

We split this skill into the rules it gives an agent

3 rules. About 0 are common, 0 are distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Read the numbers first: Run timing (MCP), or timing { tool, hours } for one tool.

    line 10
  2. Very specific:

    Go deeper only when a span can't explain it: Run py-spy record --native --pid <host pid> on the Python host, or bun --cpu-prof on Bun, to …

    line 21
  3. Very specific:

    Fix, then measure again: Run the same timing report after the change and give the owner both numbers.

    line 22

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

callTiming

The loop

  1. Read the numbers first. Run timing (MCP), or timing { tool, hours } for one tool. Note the tool's p50, p95 and p99, and the stages with the largest share.
  2. Open a slow call. Take a call id from a ■ line in the live log, or pick a slow call, and run timing { call }. The waterfall shows which stage took the time.
  3. Name the cause from the span.
    • python.start or python.imports: the cost of starting Python for each call.
    • host.connect with started=true: the account's session host was started cold.
    • host.queue or account.lock: another call held the account.
    • spacing or governor: pacing, which is there on purpose.
    • inbox.waitCaughtUp: the account was behind.
    • voice.* or fetch …: SubQ, Jina or Twenty.
    • http …: the platform itself.
    • unaccounted: our own code with no span yet. Add one.
  4. Go deeper only when a span can't explain it. Run py-spy record --native --pid <host pid> on the Python host, or bun --cpu-prof on Bun, to see which function and library the time is in.
  5. Fix, then measure again. Run the same timing report after the change and give the owner both numbers. Never call something faster without them.

Hard boundaries

  • Timing never changes what a call sends. A span holds no cookie, token or message text.
  • Pacing (spacing, governor) protects the accounts. It is never cut to win a benchmark unless the owner says so.
  • Commands go through the MCP (rule 163). timing is read-only.

Docs: docs/callTiming.md. Spec: docs/specs/callTiming.md.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free