Show HN: Daemons – we pivoted from building agents to cleaning up after them
TL;DR Highlight
DaemonMD automatically manages operational debt from AI-accelerated code generation with a single Markdown file.
Who Should Read
Development team leads and DevOps engineers who’ve adopted AI Agents (Cursor, Copilot, etc.) and are now struggling with operational overhead like abandoned PRs, documentation inconsistencies, and unclassified issues.
Core Mechanics
- Unlike Agent-driven tasks initiated by humans, Daemons operate autonomously, monitoring the environment and triggering actions when changes are detected—crucially, they function without explicit prompts.
- Daemons are defined within a repository using a single .md file, declaring their name, purpose, watch conditions, routines, prohibited actions, and execution schedule (in cron format) within the YAML frontmatter.
- The Markdown body below the frontmatter details the Daemon’s behavior in natural language, outlining operational policies, output formats, escalation rules, and execution limits—allowing role definition without code.
- Daemon behavior is explicitly restricted using ‘deny’ rules. For example, an issue-labeler Daemon might be permitted to add labels but prohibited from removing them, changing issue status, or adding comments.
- The ‘limits’ section restricts processing volume per execution. The example issue-labeler limits processing to the triggering issue on creation events and a maximum of 20 issues during daily scheduled runs, preventing excessive automation.
- Daemons can operate in a hybrid mode by combining ‘watch’ conditions and schedules, reacting to events immediately while periodically processing missed items.
- Daemon files are designed in an open format, enabling portability across providers supporting the specification, facilitating team sharing and library creation.
- Officially provided Daemon libraries include tools for maintaining PR review readiness (pr-helper), issue labeling (issue-labeler), bug triage, dependency updates (Codebase Maintainer), and documentation synchronization (Librarian).
Evidence
- "A question arose regarding collision handling when two Daemons simultaneously modify related files. The developer of agent-coordinator shared their experience implementing atomic operation preemption using SQLite’s INSERT...SELECT and assigning separate git worktrees to each agent to prevent conflicts from reaching the shared branch. The community wondered whether Daemons would lean towards additive-only behavior or explicit ordering declared in the .md file.\n\nA question compared Daemons to Claude’s Hooks feature. A comment clarified that Hooks execute once per event, while Daemons are persistent processes maintaining state across multiple events—analogous to the difference between cron and a constantly running service. Daemon’s stateful monitoring across events is suitable when single-event triggers are insufficient.\n\nMultiple comments criticized the lack of explanation regarding how Daemons integrate with existing workflows on the website. One user complained about being repeatedly told to ‘talk to Charlie’ without guidance on initiating the conversation, while another noted the absence of explanations regarding execution timing. The team responded by sharing example Daemon files and reference documentation.\n\nQuestions comparing Daemons to competing products like OpenProse were also raised, along with skepticism about whether callable skills could achieve similar results. One comment pointed out that only the schedule component is deterministic, while the rest is entirely non-deterministic.\n\nA question asked whether connecting the repository to a cloud platform would trigger the platform to read and execute the Daemon files. A clear answer wasn’t provided in the comments, with the community identifying a lack of official documentation regarding the operational architecture."
How to Apply
- "If PR review quality is inconsistent and PRs frequently lack adequate descriptions, deploy the pr-helper Daemon to automatically suggest improvements and flag missing reviewer context when PRs are opened or updated.\n\nIf Linear or GitHub Issues accumulate unlabeled issues, create an issue-labeler DAEMON.md file combining issue creation events with a daily scheduled run, and use deny rules to allow only label addition, ensuring safe automation without altering existing data.\n\nLeverage the open format of Daemon files to template operational automation by version controlling shared Daemons in a .agents/daemons/ directory and copying them for reuse in new projects.\n\nGiven the current lack of transparency regarding Daemon collision handling, narrowly define deny rules for each Daemon or design them with an additive-only approach to prevent conflicts."
Code Example
# .agents/daemons/issue-labeler/DAEMON.md example
---
name: issue-labeler
purpose: Ensures every Linear issue has the correct labels from the type and touchpoint label groups.
watch:
- when a Linear issue is created
routines:
- add missing labels to a new Linear issue
- find issues with missing labels and add them
deny:
- remove labels from issues
- replace or change existing labels on issues
- comment on issues
- change issue status, priority, assignee, or any field other than labels
schedule: "0 2 * * *"
---
## Policy
- Only add labels. Never remove, replace, or overwrite existing labels.
- If an issue already has a label from a group, do not touch that group.
- Apply the single best-fit label from each missing group.
## Limits
- On issue-created events, process only the triggering issue.
- On the daily sweep, label at most 20 issues per activation.
---
# .agents/daemons/pr-helper/DAEMON.md example
---
name: pr-helper
purpose: Keeps PRs review-ready.
watch:
- when a pull request is opened
- when a pull request is synchronized
routines:
- suggest PR description improvements
- flag missing reviewer context
deny:
- merge pull requests
- push to protected branches
schedule: "0 9 * * *"
---
## Policy
Focus on short, actionable feedback.
## Output format
1. Findings
2. Suggested edits
3. Questions for authorTerminology
Related Papers
Show HN: adamsreview – better multi-agent PR reviews for Claude Code
Claude Code에서 최대 7개의 병렬 서브 에이전트가 각각 다른 관점으로 PR을 리뷰하고, 자동 수정까지 해주는 오픈소스 플러그인이다. 기존 /review나 CodeRabbit보다 실제 버그를 더 많이 잡는다고 주장하지만 커뮤니티에서는 복잡도와 실효성에 대한 회의론도 나왔다.
How Fast Does Claude, Acting as a User Space IP Stack, Respond to Pings?
Claude Code에게 IP 패킷을 직접 파싱하고 ICMP echo reply를 구성하도록 시켜서 실제로 ping에 응답하게 만든 실험으로, 'Markdown이 곧 코드이고 LLM이 프로세서'라는 아이디어를 네트워크 스택 수준까지 밀어붙인 재미있는 사례다.
Show HN: Git for AI Agents
AI 코딩 에이전트(Claude Code 등)가 수행한 모든 툴 호출을 자동으로 추적하고, 어떤 프롬프트가 어느 코드 줄을 작성했는지 blame까지 가능한 버전 관리 도구다.
Principles for agent-native CLIs
AI 에이전트가 CLI 도구를 더 잘 사용할 수 있도록 설계하는 원칙들을 정리한 글로, 에이전트가 CLI를 도구로 활용하는 빈도가 높아지면서 이 설계 방식이 실용적으로 중요해지고 있다.
Agent-harness-kit scaffolding for multi-agent workflows (MCP, provider-agnostic)
여러 AI 에이전트가 서로 역할을 나눠 협업할 수 있도록 조율하는 scaffolding 도구로, Vite처럼 설정 없이 빠르게 멀티 에이전트 파이프라인을 구성할 수 있다.
Show HN: Tilde.run – Agent sandbox with a transactional, versioned filesystem
AI 에이전트가 실제 프로덕션 데이터를 건드려도 롤백할 수 있는 격리된 샌드박스 환경을 제공하는 도구로, GitHub/S3/Google Drive를 하나의 버전 관리 파일시스템으로 묶어준다.