How long does the integration take?

Change one base URL in an OpenAI-compatible client, or install the Python or TypeScript SDK; the quickstart routes a first request in a few minutes.

Will this lock us into another vendor?

No. The router is Apache-2.0 open source, your provider keys stay yours, and the API is OpenAI-compatible, so leaving is a change of base URL.

What happens when a provider fails?

The router falls back to the next model your policy allows and records the fallback in the trace, so an outage costs a retry, not an incident.

learn - Prompt - OpenSmartRoute

Name: learn
Author: Pranav-Shivam

Prompt file imported from Pranav-Shivam/CodingAgents (.claude/commands/learn.md). Copyright stays with the author.

Capture session learnings to docs/gotchas.md. Only extract what future Claude sessions won't know from reading the code.

Steps:

Read docs/gotchas.md — know what's already captured
Scan this session for:
- Non-obvious bugs discovered and why they happened
- Tool/command quirks (flags, platform differences, version issues)
- Agent behavior patterns that surprised you
- Workarounds for specific constraints
- Anything a new Claude session would get wrong without this note
Filter ruthlessly: skip anything derivable from the code or docs

For each lesson worth saving, write one compact entry:

## <topic>

- <specific, actionable fact> — <why it matters or when it bites you>

Output: proposed additions only. Present them. Ask user to approve before writing.

After approval: prepend entries to docs/gotchas.md under the  comment.

Rules:

No generic best practices — only project/tool-specific observations
No "remember to X" — only specific facts with concrete consequences
If nothing new discovered: say "Nothing new to capture."

Capture session learnings to docs/gotchas.md. Only extract what future Claude sessions won't know from reading the code. Steps: 1. Read `docs/gotchas.md` — know what's already captured 2. Scan this session for: - Non-obvious bugs discovered and why they happened - Tool/command quirks (flags, platform differences, version issues) - Agent behavior patterns that surprised you - Workarounds for specific constraints - Anything a new Claude session would get wrong without this note 3. Filter ruthlessly: skip anything derivable from the code or docs For each lesson worth saving, write one compact entry: ``` ## <topic> - <specific, actionable fact> — <why it matters or when it bites you> ``` Output: proposed additions only. Present them. Ask user to approve before writing. After approval: prepend entries to docs/gotchas.md under the `` comment. Rules: - No generic best practices — only project/tool-specific observations - No "remember to X" — only specific facts with concrete consequences - If nothing new discovered: say "Nothing new to capture."

Install learn and it becomes one more candidate the router can pick - when it fits.

A listing is a routing target with a manifest: what it does, which domains it covers, what it costs and who publishes it. Once installed it sits beside your own models and tools, is scored like any other candidate for each request, and shows up in the trace when it wins. Ratings come from workspaces that installed it, one per account.

{ "prompt": "Capture session learnings to docs/gotchas.md. Only extract what future Claude sessions won't know from reading the code.\n\nSteps:\n1. Read `docs/gotchas.md` — know what's already captured\n2. Scan this session for:\n - Non-obvious bugs discovered and why they happened\n - Tool/command quirks (flags, platform differences, version issues)\n - Agent behavior patterns that surprised you\n - Workarounds for specific constraints\n - Anything a new Claude session would get wrong without this note\n3. Filter ruthlessly: skip anything derivable from the code or docs\n\nFor each lesson worth saving, write one compact entry:\n```\n## <topic>\n\n- <specific, actionable fact> — <why it matters or when it bites you>\n```\n\nOutput: proposed additions only. Present them. Ask user to approve before writing.\n\nAfter approval: prepend entries to docs/gotchas.md under the `` comment.\n\nRules:\n- No generic best practices — only project/tool-specific observations\n- No \"remember to X\" — only specific facts with concrete consequences\n- If nothing new discovered: say \"Nothing new to capture.\"", "variables": [], "notes": "Extract non-obvious learnings from this session and save to docs/gotchas.md. Invoke with /learn at session end. Never auto-triggers. Tools: Read, Bash(git diff HEAD), Bash(git log --oneline -5).", "metadata": { "source": { "provider": "github-claude-commands", "repository": "https://github.com/Pranav-Shivam/CodingAgents", "path": ".claude/commands/learn.md", "ref": "2c1227c8c628328b2d4d0a0ce277852d5a9fe3c1", "url": "https://github.com/Pranav-Shivam/CodingAgents/blob/2c1227c8c628328b2d4d0a0ce277852d5a9fe3c1/.claude/commands/learn.md", "key": "Pranav-Shivam/CodingAgents/.claude/commands/learn.md" } }, "model_hints": [ "haiku" ] }

learn

About

Use it

Install learn and it becomes one more candidate the router can pick - when it fits.

What it costs to adopt - in engineering time.

One base URL

No vendor lock-in

Failover built in

Manifest

Reviews