Glean Performance Tuning

What it does

When your agent uses it

• Tasks that involve Search implementation

Example prompts

• “glean performance”
• “glean search quality”
• “glean indexing speed”
• “/glean-performance-tuning”

Requirements

• Compatibility (from SKILL.md): Designed for Claude Code
• Pre-approved tools (allowed-tools): Read, Write, Edit

Workflow steps

5 steps, taken from the first numbered list in SKILL.md.

1. 1Establish a baseline and target one bottleneck at a time: source read, transform, index submission, permission sync, or query.
2. 2Test bounded batch and concurrency changes in staging with backpressure, idempotency, and strict retry limits.
3. 3Confirm cache invalidation after ACL and document updates so lower latency never serves unauthorized or stale results.
4. 4Canary one datasource, monitor latency, freshness, rate limits, and synthetic allow/deny probes, then promote or roll back.
5. 5Record the configuration revision and review cost, error budget, and authorization evidence together.

What it can do on your machine

Read from SKILL.md and the folder at commit cfae287. It shows what the files ask for, not the result of running them.

• Tool permissions

  Pre-approves these tools, so the agent can use them without asking each time:

  • Read
  • Write
  • Edit

  From allowed-tools in the SKILL.md frontmatter.

• Runs code

  No scripts in the folder and no shell commands in SKILL.md (its code samples are typescript).

  From the folder's file list and the shell code blocks in SKILL.md.

• Network

  Links to these hosts (documentation or services it may open):

  • developers.glean.com

  From URLs in SKILL.md, links to its own repository left out.

• Credentials

  Names no API keys, tokens, secrets or passwords.

  From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

• Compatibility

  Designed for Claude Code

  From compatibility in the SKILL.md frontmatter.

Context cost

Glean Performance Tuning loads about 1.5k tokens when it runs. Until then it costs about 59 tokens; SKILL.md has 428 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it

~59

When it runs · the whole SKILL.md, loaded when a task matches

~1.5k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from jeremylongshore/tons-of-skills-marketplace at commit cfae287, republished under its MIT licence (© jeremylongshore). 428 words, ~1,464 tokens.

name

glean-performance-tuning

description

Optimize Glean search relevance and indexing throughput with batch sizing, datasource configuration, and content quality improvements. Trigger: "glean performance", "glean search quality", "glean indexing speed".

allowed-tools

Read, Write, Edit

compatibility

Designed for Claude Code

version

1.8.0

license

MIT

author

Jeremy Longshore <jeremy@intentsolutions.io>

tags

saas, enterprise-search, glean

const cache = new Map<string, { data: any; expiry: number }>();
const TTL = { search: 60_000, suggestions: 30_000, datasources: 600_000 };

async function cached(key: string, ttlKey: keyof typeof TTL, fn: () => Promise<any>) {
  const entry = cache.get(key);
  if (entry && entry.expiry > Date.now()) return entry.data;
  const data = await fn();
  cache.set(key, { data, expiry: Date.now() + TTL[ttlKey] });
  return data;
}
// Search results expire fast (1 min). Datasource metadata is stable (10 min).

import PQueue from 'p-queue';
const BATCH_SIZE = 100;

async function indexDocsBatched(glean: any, dsName: string, docs: any[]) {
  const batches = [];
  for (let i = 0; i < docs.length; i += BATCH_SIZE) batches.push(docs.slice(i, i + BATCH_SIZE));
  const queue = new PQueue({ concurrency: 3, interval: 500 });
  await Promise.all(batches.map(batch =>
    queue.add(() => glean.indexDocuments(dsName, batch))
  ));
}

import { Agent } from 'https';
const agent = new Agent({ keepAlive: true, maxSockets: 15, maxFreeSockets: 5, timeout: 30_000 });
// High socket count for parallel indexing across multiple datasources

async function withGleanRateLimit(fn: () => Promise<any>): Promise<any> {
  try { return await fn(); }
  catch (err: any) {
    if (err.status === 429) {
      const retryMs = parseInt(err.headers?.['retry-after'] || '5') * 1000;
      await new Promise(r => setTimeout(r, retryMs));
      return fn();
    }
    throw err;
  }
}

const metrics = { searches: 0, indexOps: 0, cacheHits: 0, p95LatencyMs: 0, errors: 0 };
const latencies: number[] = [];
function trackSearch(startMs: number, cached: boolean) {
  const lat = Date.now() - startMs; latencies.push(lat); metrics.searches++;
  if (cached) metrics.cacheHits++;
  latencies.sort((a, b) => a - b);
  metrics.p95LatencyMs = latencies[Math.floor(latencies.length * 0.95)] || 0;
}

© jeremylongshore, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in skills/.curated/glean-performance-tuning of jeremylongshore/tons-of-skills-marketplace.

Open the folder on GitHubat commit cfae287