# SpecSource

> Automated Bug Specifications

SpecSource sources detailed specifications for your Linear issues. AI-powered research that gathers context from Sentry, GitHub, and Slack — so your team ships fixes, not paperwork.

## The Problem

Every bug starts the same way — digging through error logs, Slack threads, and PRs just to write the spec. Engineers spend 15–30 minutes per issue investigating context before they can even start fixing. For a team handling 15–30 bugs per day, that's 15–30 engineering hours per week — the equivalent of one full-time engineer doing nothing but triage.

Context switching kills velocity. It takes 23 minutes to recover focus after an interruption, and triage disproportionately falls on senior engineers and tech leads — the people whose time is most valuable.

## How It Works

1. **Connect your stack** — Add your Linear, Sentry, GitHub, and Slack integrations. Takes about five minutes.
2. **Issues get researched** — When a new Linear issue arrives, the AI agent finds related errors, code, PRs, and Slack threads.
3. **Specs write themselves** — All that context gets compiled into a detailed specification that enriches your Linear issue.

The agent reads the Linear issue, checks Sentry for related errors and stack traces, gathers relevant source code, recent commits, and PRs from GitHub, searches Slack for related team discussions, checks for semantic duplicates using vector embeddings, then writes a structured specification synthesizing everything — all in under 30 seconds.

## Features

- **Smart issue matching** — Matches new issues to existing Linear issues using semantic similarity. No more duplicate specs.
- **Code context, included** — Pulls the relevant files, recent PRs, and commits so you see the root cause without digging.
- **Cross-tool context gathering** — The agent queries Sentry, GitHub, and Slack automatically — so you never switch tabs to build context.
- **Auto-generated specifications** — Issue data, code context, and team discussions compiled into one detailed specification, enriching your Linear issue.
- **Semantic deduplication** — Uses vector embeddings to find duplicates that keyword matching misses.
- **AI-agent-ready output** — Specifications structured so Claude, Cursor, and Copilot can act on them immediately.
- **MCP support** — Extend the agent with custom internal tools and APIs via Model Context Protocol.
- **Multiple triggers** — Run the agent via webhooks, on a schedule, or manually.

## Integrations

- **Linear** — Available now
- **Sentry** — Available now
- **GitHub** — Available now
- **Slack** — Available now
- **MCP** — Available now (extend with custom tools)
- **Jira** — Coming soon
- **Asana** — Coming soon
- **Trello** — Coming soon

## Pricing

**Free — $0 forever**
- 100 credits/month
- 1 project
- All connectors (Linear, Sentry, GitHub, Slack)
- 1 MCP server
- Standard AI models (GPT-5 Mini, Claude Haiku 4.5)
- Scheduling

**Pro — $10/month**
- 1,000 credits/month
- Unlimited projects
- All connectors including MCP
- Unlimited MCP servers
- Premium AI models (GPT-5.2, Claude Sonnet 4.5)
- Deep context gathering
- Priority support

**Pro Usage — $10/month + usage**
- $10 per 1,000 credits
- Up to 5,000 credits/month
- All Pro features
- Pay only for what you use

A typical agent run consumes 5–20 credits. No credit card required to start.

## Use Cases

- [AI Agent Context](https://www.specsource.ai/use-cases/ai-agent-context) — Automatically generate prompt-ready specifications your AI coding agent can act on immediately.
- [Automated Bug Research](https://www.specsource.ai/use-cases/automated-bug-research) — Linear issues research themselves across Sentry, your codebase, and Slack.
- [Linear Specification Sourcing](https://www.specsource.ai/use-cases/linear-ticket-generation) — Detailed specs for every Linear issue with root cause analysis and code context.
- [Bug Triage for Startups](https://www.specsource.ai/use-cases/bug-triage-for-startups) — Enterprise-grade bug triage for lean engineering teams.
- [AI Bug Fixing with Cursor & Claude](https://www.specsource.ai/use-cases/cursor-claude-bug-fixing) — Write bug specs your AI coding agent actually needs to fix issues.
- [Engineering Team Productivity](https://www.specsource.ai/use-cases/engineering-team-productivity) — Give your engineers their time back by automating bug triage.

## FAQ

**What are credits?**
Credits are the currency for AI agent runs. Each run consumes credits based on the model used and tokens processed. Free plan includes 100 credits/month, Pro includes 1,000.

**Which AI models are included?**
Free plan uses GPT-5 Mini and Claude Haiku 4.5. Pro plan unlocks GPT-5.2 and Claude Sonnet 4.5 for higher quality specifications.

**Can I switch plans anytime?**
Yes. Upgrade to Pro instantly from dashboard settings. Downgrade anytime — you keep Pro features until the end of your billing period.

## Company

SpecSource is operated by Quassum MB, based in Lithuania.

## Links

- [Website](https://www.specsource.ai)
- [Documentation](https://www.specsource.ai/docs)
- [Blog](https://www.specsource.ai/blog)
- [Getting Started Video](https://www.youtube.com/watch?v=bTnoaAxv2Sw)
- [Get Started Free](https://www.specsource.ai/login)
- [Privacy Policy](https://www.specsource.ai/privacy)
- [Terms of Service](https://www.specsource.ai/terms)