What is improve-agent?

A command for improving an Algolia Agent Studio agent by evaluating a safe copy, examining failures, testing setting changes, and promoting a validated version only after approval.

It is for running agent evaluations, investigating failed cases, comparing draft changes, and improving an agent's evaluation score.

It avoids changing the active agent before comparing results, and makes the evaluation and approval steps explicit.

improve-agent — Command for Claude Code by algolia/skills

Install

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.

Clone the repo

git clone --depth 1 https://github.com/algolia/skills

Made for: Claude Code.

Wrote 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.

README.md

[![agentmods](https://agentmods.dev/badge/commands/algolia/skills/improve-agent/github.svg)](https://agentmods.dev/commands/algolia/skills/improve-agent)

Your own site

<a href="https://agentmods.dev/commands/algolia/skills/improve-agent"><img src="https://agentmods.dev/badge/commands/algolia/skills/improve-agent/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.

Your own site · 80×15

<a href="https://agentmods.dev/commands/algolia/skills/improve-agent"><img src="https://agentmods.dev/badge/commands/algolia/skills/improve-agent.svg" alt="Reviewed on agentmods" width="80" height="20"></a>

Per session 26 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.

When invoked 4,454 The whole file, excluding the scripts and references it only reads on demand.

Security scan A 0 findings. A grade says what 26 rules found in the file — not that it is safe.

Origin original No closer match found in the catalogue.

Token cost

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	$0.00026	$0.04454
Opus 5.5	$0.00010	$0.01782
Sonnet 5.5	$0.00005	$0.00891
Haiku 4.5	$0.00003	$0.00445

Measured 11d ago against content hash 33e2615b7bdc, method: parsed. Prices are Anthropic first-party input rates as of 2026-10-07, from the pricing page.

Security

Grade A, and why

improve-agent 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 11d 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.

commands/improve-agent.md · 346 lines

How it starts

The opening of the file, as written. The whole thing — 346 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Improve Agent (eval-driven tuning)

Improve an Agent Studio agent's eval score: run an evaluation, analyze failing cases, change agent settings on a draft clone, re-evaluate, compare, and promote only with the user's explicit approval.

Load the agent-studio-evaluations skill (and references/evaluations.md inside it) before running any command below. For pure CLI mechanics (install, auth, agents/chat/profiles command syntax) also see the algobot-cli skill.

SKILL_DIR below means the directory containing the agent-studio-evaluations skill's SKILL.md (wherever your agent installed it). Its helper scripts live in $SKILL_DIR/scripts/ and its templates in $SKILL_DIR/templates/. Thresholds (headroom, sample size, cost, promotion gate) are defined once in references/evaluations.md; this command points to them.

Step 0: Preconditions

Confirm the CLI has evals support: algobot evals metrics list. Evaluations require algobot-ai >= 2.1.0 (npm latest). If the command is unknown, the installed version predates evals: stop and ask the user to upgrade (npm i -g algobot-ai@latest) and verify with algobot --version.
Resolve the target agent: algobot agents list / algobot agents get <id>. Record: id, name, status, providerId, index (tools[].indices[]).
Ask the user for metrics if not specified (algobot evals metrics list), a total case count (the CLI default is 25; e.g. 300 for 3 metrics x 100 each), and any focus (tone, faithfulness, ...).
Announce the cost: every case = a real completion + judge calls. --num-cases is a TOTAL split across the selected metrics, not a per-metric count (observed: 100 with 2 metrics gave 50 each). For 3 metrics x 100 cases each, pass --num-cases 300 (about 300 completions and 300 judge calls). Imported dataset: one metric per case.
Run the zero-cost static audit first (commands/audit-agent.md) and fix blockers before spending on evals.
Dataset gate: every run you will compare passes --dataset-id. Before each paid run, print a run card ($SKILL_DIR/templates/run-card.md): dataset id and name, agentProfileId, indexProfileId, cases per metric, distinct tone contracts (tone_constraint + specific_rule). Observed: experiments create without --dataset-id resolved to an existing dataset for the same index built for a DIFFERENT agent, so a second agent inherited the first agent's tone contract. Refuse to compare runs from different datasets.
Take the rollback snapshot now, before any tuning, so the promotion gate can name it, and start a lineage ledger:

Read the full file on GitHub · 346 lines

Changes

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.

11d ago Changed · +156 lines 33e2615b7bdc
14d ago First seen · 190 lines · 26 tokens per session scan A bee41200f14f

Subscribe to this mod's changes

improve-agent is a command published in the GitHub repository algolia/skills (12 stars, last pushed 7d ago), licensed MIT. It adds 26 tokens to every session and 4,454 once invoked, about $0.0001 per session on Opus 5.5. 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 2026-10-24.

Analytics and metrics Building agents agent ai-assistant algolia mcp recommendations search

improve-agent next to 6 commands from other repositories
Mod	Type	Stars	Last push	Licence	Scan	Tokens	Works with
improve-agentalgolia/skills	Command	12	7d ago	MIT	A	26	Claude Code
screenshotStarTrail-org/PixelRAG	Command	11k	yesterday	Apache-2.0	A	12	Claude Code
checklistgithub/spec-kit	Command	148k	yesterday	MIT	A	13	Claude Code
clarifygithub/spec-kit	Command	148k	yesterday	MIT	A	27	Claude Code
specifygithub/spec-kit	Command	148k	yesterday	MIT	A	13	Claude Code
analyzegithub/spec-kit	Command	148k	yesterday	MIT	A	26	Claude Code
convergegithub/spec-kit	Command	148k	yesterday	MIT	A	35	Claude Code