write-continue-here
À propos
Cette compétence Claude crée un fichier CONTINUE_HERE.md pour préserver le contexte des sessions de codage, permettant des transferts fluides entre sessions. Elle structure le fichier avec des sections objectif, suivi de progression et prochaines étapes lors de l'interruption d'un travail inachevé. La compétence garantit que ce fichier de continuation est exploitable pour de nouvelles sessions Claude Code, là où git seul ne peut capturer l'état de la tâche.
Installation rapide
Claude Code
Recommandénpx skills add pjt222/agent-almanac -a claude-code/plugin add https://github.com/pjt222/agent-almanacgit clone https://github.com/pjt222/agent-almanac.git ~/.claude/skills/write-continue-hereCopiez et collez cette commande dans Claude Code pour installer cette compétence
Documentation
Write Continue Here
Write a structured continuation file so the next session starts with full context.
When to Use
- Ending a session with work still in progress
- Handing off a complex task between sessions
- Preserving intent, failed approaches, and next steps that git cannot capture
- Before closing Claude Code when mid-task
Inputs
- Required: An active session with recent work to summarize
- Optional: Specific instructions about what to emphasize in the handoff
Procedure
Step 1: Assess Session State
Gather facts about recent work:
git log --oneline -5
git status
git diff --stat
Review the conversation context: what was the objective, what was completed, what is partially done, what was tried and failed, what decisions were made.
Got: Clear understanding of current task state — completed items, in-progress items, and planned next steps.
If fail: If not in a git repository, skip git commands. The continuation file can still capture conversational context and task state.
Step 2: Write CONTINUE_HERE.md
Write the file to the project root using the structure below. Every section must contain actionable content, not placeholders.
# Continue Here
> Last updated: YYYY-MM-DDTHH:MM:SSZ | Branch: current-branch-name
## Objective
One-paragraph description of what we are trying to accomplish and why.
## Completed
- [x] Finished item with key file paths (e.g., `src/feature.R`)
- [x] Decisions made and their rationale
## In Progress
- [ ] Partially complete work — describe current state (branch, file:line)
- [ ] Known issues with partial work
## Next Steps
1. Immediate next action (most important)
2. Subsequent actions in priority order
3. **[USER]** Items needing user input or decision
## Context
- Failed approaches and why they did not work
- Key constraints or trade-offs discovered
- Relevant issue/PR links
Guidelines:
- Objective: Capture the WHY — git log shows what changed, not why
- Completed: Mark items clearly done to prevent re-work
- In Progress: This is the highest-value section — partial state is hardest to reconstruct
- Next Steps: Number by priority. Prefix user-dependent items with
**[USER]** - Context: Record negative space — what was tried and rejected, and why
Got: A CONTINUE_HERE.md file at the project root with all 5 sections populated with real content from the current session. The timestamp and branch are accurate.
If fail: If Write fails, check file permissions. The file should be created in the project root (same directory as .git/). Verify .gitignore contains CONTINUE_HERE.md — if not, add it.
Step 3: Verify the File
Read back CONTINUE_HERE.md and confirm:
- Timestamp is current (within the last few minutes)
- Branch name matches
git branch --show-current - All 5 sections contain real content (no template placeholders)
- Next Steps are numbered and actionable
- In Progress items describe current state specifically enough to resume
Got: The file reads as a clear, actionable handoff that a fresh session could use to immediately resume work.
If fail: Edit sections that contain placeholder text or are too vague. Each section should pass the test: "Could a fresh session act on this without asking clarifying questions?"
Validation
- CONTINUE_HERE.md exists at the project root
- File contains all 5 sections with real content (not placeholders)
- Timestamp and branch are accurate
-
.gitignoreincludesCONTINUE_HERE.md - Next Steps are numbered and actionable
- In Progress items specify enough detail to resume without questions
Pitfalls
- Writing placeholders instead of content: "TODO: fill in later" defeats the purpose. Every section must contain real information from the current session.
- Duplicating git state: Do not list every file changed — git already tracks that. Focus on intent, partial state, and next steps.
- Forgetting the Context section: Failed approaches are the most valuable thing to record. Without them, the next session will retry the same dead ends.
- Overwriting without reading: If CONTINUE_HERE.md already exists from a prior session, read it first — it may contain unfinished work from an earlier handoff.
- Leaving stale files: CONTINUE_HERE.md is ephemeral. After the next session consumes it, delete it. Stale files cause confusion.
Related Skills
read-continue-here— the complement: reading and acting on the continuation file at session startbootstrap-agent-identity— cold-start identity reconstruction that consumes the continuation file this skill producesmanage-memory— durable cross-session knowledge (complements this ephemeral handoff)commit-changes— save work to git before writing the continuation filewrite-claude-md— project instructions where optional continuity guidance lives
Dépôt GitHub
Compétences associées
railway-docs
DocumentationCette compétence récupère la documentation actuelle de Railway pour répondre aux questions sur les fonctionnalités, le fonctionnement ou des URL spécifiques de la documentation. Elle garantit que les développeurs reçoivent des informations précises et à jour directement depuis les sources officielles de Railway. Utilisez-la lorsque les utilisateurs demandent comment fonctionne Railway ou font référence à la documentation de Railway.
n8n-code-python
DocumentationCette compétence Claude offre un accompagnement expert pour écrire du code Python dans les nœuds Code de n8n, en particulier pour utiliser la bibliothèque standard de Python et travailler avec la syntaxe spéciale de n8n comme `_input`, `_json` et `_node`. Elle aide les développeurs à comprendre les limites de Python dans n8n et recommande d'utiliser JavaScript pour la plupart des workflows, tout en proposant des solutions Python pour des besoins spécifiques de transformation de données.
archon
DocumentationLa compétence Archon offre une recherche sémantique alimentée par RAG et une gestion de projet via une API REST. Utilisez-la pour interroger la documentation, gérer des projets/tâches hiérarchiques et effectuer de la recherche de connaissances avec des capacités de téléchargement de documents. Priorisez toujours Archon en premier lors de la recherche dans une documentation externe avant d'utiliser d'autres sources.
n8n-code-javascript
DocumentationCette compétence Claude fournit des conseils d'expert pour écrire du code JavaScript dans les nœuds Code d'n8n. Elle couvre la syntaxe essentielle spécifique à n8n comme les variables `$input`/`$json`, les assistants HTTP et la gestion des DateTime, tout en résolvant les erreurs courantes. Utilisez-la lors du développement de workflows n8n nécessitant un traitement JavaScript personnalisé dans les nœuds Code.
