orchestrator-prompt-editing
This skill is currently disabled and non-functional because it targets an outdated system prompt file and incorrect XML structure.
Install
mkdir -p .claude/skills/orchestrator-prompt-editing && curl -L -o skill.zip "https://agentskills.codes/api/skills/download/14462" && unzip -o skill.zip -d .claude/skills/orchestrator-prompt-editing && rm skill.zipInstalls to .claude/skills/orchestrator-prompt-editing
Activation
This is the description your AI agent reads to decide when to run this skill — the better it matches your request, the more reliably it fires.
[STALE — DO NOT USE] Skill is disabled pending refresh. It targets the production prompt path/structure, but redesign mode swapped to a different file with a different structure on 2026-04-27. Edit prompts manually until this skill is updated.Key capabilities
- →Identify the active orchestrator prompt file.
- →Document the structure of orchestrator prompts.
- →Add directives and constraints to the prompt XML file.
- →Validate XML file structure.
- →Propose changes via a `question-for-eva` issue.
How it works
This skill is intended to edit the orchestrator's system prompt by identifying the active prompt file, reading journal entries for behavioral issues, and then adding directives or constraints to the XML file, followed by validation.
Inputs & outputs
When to use orchestrator-prompt-editing
- →Manually editing system prompt XML files
- →Updating prompt structure after architectural changes
- →Refining phase orderings and checkpoints
- →Clarifying prompt scope and mission statements
About this skill
[STALE — DO NOT USE] Editing the Orchestrator System Prompt
This skill is disabled as of 2026-04-29. Three known staleness issues:
- Wrong file targeted. Skill assumes
.github/workflows/orchestrator-prompt.xml, but redesign mode (installed 2026-04-27, commit06c799cd) swapped the active prompt to.github/workflows/orchestrator-redesign-prompt.xmlvia.github/workflows/orchestrator.ymlline 51. Both files exist; only one is active. The skill predates this swap.- Wrong XML structure. Skill describes the production prompt's 8 sections (
<identity>,<directives>,<definitions>,<environment>, three<phase>sections,<practices>). The redesign prompt has a completely different structure:<critical-context-read-first>,<mission>,<core-design-principle>,<authority>,<phases>,<checkpoints>,<abort-criteria>,<initial-directive>,<persistence>,<iteration-until-approval>, etc.- "When to edit" framing is production-only. Skill triggers are all behavioral-correction signals (recurring findings, adoption gaps, data-filtering). Redesign-mode edits are usually design-refinement (phase ordering, checkpoint tweaks, scope clarifications) — different shape entirely.
To make current: detect which prompt is active (read
.github/workflows/orchestrator.ymland find thecatline); document both prompt structures; add a redesign-mode editing subsection. Until then, edit prompts manually with surgical edits and validate XML afterward.
Editing the Orchestrator System Prompt
The orchestrator's system prompt lives at .github/workflows/orchestrator-prompt.xml. This XML file is loaded as the orchestrator's briefing every cycle. It contains ALL orchestrator instructions — identity, directives, definitions, environment, startup steps, close-out steps, and standing practices. Changes here directly affect orchestrator behavior.
Key facts
- Location:
.github/workflows/orchestrator-prompt.xml— a single XML file that replaces the previous markdown prompt + STARTUP_CHECKLIST.md + COMPLETION_CHECKLIST.md - Format: XML with 8 top-level sections:
<identity>,<directives>,<definitions>,<environment>,<phase id="startup">,<phase id="work">,<phase id="close-out">,<practices> - The workflow YAML (
.github/workflows/orchestrator.yml) loads this file viacatand passes it as the system prompt - Human-only change: The orchestrator cannot modify this file itself (it's under
.github/workflows/). It must create aquestion-for-evaissue for proposed changes.
When to edit
Edit the prompt when journal entries or review findings reveal recurring behavioral patterns that tools haven't fixed. Indicators:
- The same finding category appears across 3+ consecutive review cycles
- The orchestrator acknowledges a pattern in its journal but doesn't change behavior
- Tools exist but aren't being used (adoption gaps)
- The orchestrator interprets or filters data instead of reporting canonical values
How to edit
- Read the journal entries that document the behavioral issue
- Read the current prompt at
.github/workflows/orchestrator-prompt.xml - Edit the XML file — add directives/constraints in the appropriate section
- Validate:
python3 -c "import xml.etree.ElementTree as ET; ET.parse('.github/workflows/orchestrator-prompt.xml')" - Commit — Eva will need to merge via PR (orchestrator has no Workflows permission)
XML structure
| Section | Purpose | When to edit |
|---|---|---|
<directives> | Non-negotiable behavioral constraints | Recurring behavioral failures |
<definitions> | Enumerations, thresholds, labels, tools | New thresholds, finding categories, tools |
<environment> | Runtime, permissions, communication | Infrastructure changes |
<phase id="startup"> | Steps S0-S9 | Startup procedure changes |
<phase id="close-out"> | Steps C1-C8 | Close-out procedure changes |
<practices> | Standing reference (worklog, journal, tools) | Practice updates |
Behavioral directives (<directives> section) is the right place for constraints addressing recurring failures. Each directive should have id, severity="mandatory", <rule>, and optionally <violation-signal>.
What NOT to do
- Do not let the orchestrator self-modify its prompt — requires human oversight
- Do not duplicate content — each rule should be defined once. Use
ref-directive,ref-threshold, andcross-refattributes for references - Do not break XML validity — always validate after editing
When not to use it
- →When the skill is disabled due to staleness issues.
- →When the orchestrator needs to self-modify its prompt.
- →When the user needs to duplicate content within the prompt.
Limitations
- →The skill is currently disabled due to targeting the wrong file and having an outdated XML structure.
- →The orchestrator cannot self-modify its prompt.
- →It prohibits duplicating content within the prompt.
How it compares
This skill provides a structured approach to modifying the orchestrator's system prompt based on observed behavioral patterns, which is more targeted than general prompt engineering or manual trial-and-error.
Compared to similar skills
orchestrator-prompt-editing side by side with the closest alternatives in the catalog.
| Skill | Installs | Updated | Safety | Difficulty |
|---|---|---|---|---|
| orchestrator-prompt-editing (this skill) | 0 | 3mo | No flags | Advanced |
| prompt-optimizer | 43 | 6mo | No flags | Beginner |
| prompt-optimize | 13 | 9mo | No flags | Advanced |
| ai-cost-optimizer | 9 | 5mo | Caution | Intermediate |
Try saying
Example prompts that trigger this skill in your AI assistant.
You might also like
prompt-optimizer
solatis
Optimize system prompts for Claude Code agents using proven prompt engineering patterns. Use when users request prompt improvement, optimization, or refinement for agent workflows, tool instructions, or system behaviors.
prompt-optimize
YYH211
Expert prompt engineering skill that transforms Claude into "Alpha-Prompt" - a master prompt engineer who collaboratively crafts high-quality prompts through flexible dialogue. Activates when user asks to "optimize prompt", "improve system instruction", "enhance AI instruction", or mentions prompt engineering tasks.
ai-cost-optimizer
ScientiaCapital
Save 40-70% on AI costs with intelligent multi-LLM routing. Automatically selects the optimal model based on task complexity across 40+ models from 8 providers.
prompt-library
davila7
Curated collection of high-quality prompts for various use cases. Includes role-based prompts, task-specific templates, and prompt refinement techniques. Use when user needs prompt templates, role-play prompts, or ready-to-use prompt examples for coding, writing, analysis, or creative tasks.
context-compression
muratcankoylan
This skill should be used when the user asks to "compress context", "summarize conversation history", "implement compaction", "reduce token usage", or mentions context compression, structured summarization, tokens-per-task optimization, or long-running agent sessions exceeding context limits.
prompt-lookup
f
Activates when the user asks about AI prompts, needs prompt templates, wants to search for prompts, or mentions prompts.chat. Use for discovering, retrieving, and improving prompts.