Soul HaulA field guide to agent souls

The files people write to give AI agents a personality, collected from the public web every day.

The daily 5Submit a soul
  • Index911
  • New this week
  • Saved

Kinds

  • Soul879
  • Agent rules9
  • Persona18
  • Bot profile11
  • Framework config3

Archetypes

all
  • Blunt expert148
  • Friendly helper44
  • Coding pair109
  • Research analyst116
  • Creative writer7
  • Companion / character101
  • Personal clone3
  • Business operator336
  • Teacher / coach20
  • Experimental / weird26

Agents

  • OCOpenClaw762
  • HEHermes743
  • BOBrowserOS679
  • OTOther or custom110
  • CCClaude Code11
  • CXCodex5
  • CLClaude3
  • GKGrok2
  • CRCrewAI1
  • CUCursor1
AboutPrivacy
Soul HaulA field guide to agent souls

Soul Haul catalogs agent personality files from public sources. Every entry links back to its original, and authors can have theirs removed at any time.

AboutArchetypesNewsletterTakedownPrivacy
  • Index911
  • New this week
  • Saved

Kinds

  • Soul879
  • Agent rules9
  • Persona18
  • Bot profile11
  • Framework config3

Archetypes

all
  • Blunt expert148
  • Friendly helper44
  • Coding pair109
  • Research analyst116
  • Creative writer7
  • Companion / character101
  • Personal clone3
  • Business operator336
  • Teacher / coach20
  • Experimental / weird26

Agents

  • OCOpenClaw762
  • HEHermes743
  • BOBrowserOS679
  • OTOther or custom110
  • CCClaude Code11
  • CXCodex5
  • CLClaude3
  • GKGrok2
  • CRCrewAI1
  • CUCursor1

Kind

Soul

Identity and personality files such as SOUL.md, IDENTITY.md and STYLE.md.
Soul 879Agent rules 9Persona 18Bot profile 11Framework config 3

451 · № 55AD·028E · Friendly helper

★ 54

Crablet: priority-driven Chinese AI assistant with dual-system reasoning

Sets up Crablet, a general-purpose assistant that ranks user needs, honesty, efficiency and privacy as core values, separates facts from guesses, and switches between fast and deep reasoning modes depending on task complexity. Defaults to Chinese.

professionalstructuredOCHEBOSoulGitHubOct 22, 2026
View

452 · № B8C1·3C1C · Coding pair

★ 25

QA test generator agent for Cypress, Playwright and WebdriverIO

A focused QA agent that turns plain-English requirements into deterministic, CI-friendly end-to-end tests and diagnoses failures in a strict CATEGORY/REASON/FIX format. It sets rules on selectors, secrets and git safety, with no personality flourish.

terseformalOCHEBOSoulGitHubOct 22, 2026
View

453 · № 51D7·826C · Blunt expert

★ 23

Direct, privacy-minded soul file for OpenClaw-style agents

Sets up a direct, concise agent that solves problems independently, shows its reasoning, protects user privacy and credentials, and asks permission before risky or irreversible actions. The personality is mostly generic, with a few specific tone choices.

directconcisetechnicalOCHEBOSoulGitHubOct 22, 2026
View

454 · № 7039·96B9 · Blunt expert

★ 48

Repository-local engineering specialist for Omni (Hermes)

Sets up a cautious repo-local engineering agent that grounds conclusions in repository evidence, follows existing patterns, validates with static gates, and avoids live actions like external calls, daemons, database access, or production changes without approval.

terseformalcautiousHESoulGitHubOct 22, 2026
View

455 · № B67B·76F6 · Blunt expert

★ 44

Security-first trustworthy assistant with strict injection and leak guards

Sets up a pragmatic, verification-driven assistant that treats external content as untrusted, requires explicit confirmation for sensitive or destructive actions, and blocks secret leaks and suspicious scripts.

pragmaticcautiousformalOCHEBOSoulGitHubOct 21, 2026
View

456 · № 63A8·C25E · Blunt expert

Conductor persona that plans, delegates and verifies work

Sets up an orchestrator that acts as a conductor rather than a worker: it reuses prior solutions, demands verified output, pushes back on wrong framing, and spawns the minimum set of least-privilege specialists before closing every loop.

directtersepragmaticOTHESoulGitHubOct 21, 2026
View

457 · № 84B5·9E00 · Research analyst

★ 1

Analytical checklist assistant for Brazilian public procurement (Lei 14.133)

Sets up mAI, a procurement specialist for LHFEX that works from the real text of public tender notices under Lei 14.133. It highlights deadlines, documents, requirements and risks in an objective, checklist-first format, and never assumes a company is qualified without verification.

analiticoobjetivometodicoOCHEBOSoulGitHubOct 21, 2026
View

458 · № 0D22·4BC7 · Coding pair

★ 1

Direct, grounded coding agent for smolbot

Sets up smolbot as a practical coding agent that speaks calmly and directly, avoids claiming unwired capabilities, and prefers small testable changes over broad rewrites. It also tells the agent to load any skill that might apply.

directcalmpreciseOCHEBOSoulGitHubOct 22, 2026
View

459 · № D4B6·9229 · Business operator

★ 1

Ops agent with strict task, memory and bus-logging discipline

Sets up an autonomous, system-first operations agent that logs every action to the bus, creates tasks before significant work, keeps two-layer memory updated each heartbeat, and asks approval before external or destructive actions. Communication is direct and concise, and it escalates after 15 minutes stuck.

directterseproceduralCCOCHEBOSoulGitHubOct 21, 2026
View

460 · № 2235·73F1 · Friendly helper

★ 1

Genuinely helpful, opinionated assistant soul template (Korean)

A Korean SOUL.md template that sets up an assistant which acts before asking, has real opinions, avoids filler praise, treats access to a user's life as trust, and keeps its identity in editable files. It is a generic starting point with some well-crafted lines.

warmdirectconciseOCHEBOSoulGitHubOct 21, 2026
View

461 · № E704·426E · Business operator

★ 1

Strategic product manager for Discord-driven issue refinement

Sets up a structured, analytical Product Manager agent that turns demands into clear, testable GitHub Issues, refines vague scope with the user, and actively moderates a Discord channel to keep discussion on the backlog. It is barred from writing code, executing technical tasks, or approving PRs.

strategicanalyticalstructuredOCSoulGitHubOct 22, 2026
View

462 · № CAFA·DE0A · Coding pair

★ 6

Character visual-DNA locker for a cinesmith crew agent

Sets up an agent that locks a character's visual identity (face, exact eye shade, hair, costume, marks) and records it in a characters.md file so it never drifts. The style is terse and procedural.

tersepreciseOCHEBOSoulGitHubOct 21, 2026
View

463 · № 60E5·55D4 · Friendly helper

★ 149

Warm, direct, technically honest voice for a SOUL agent

Sets up a warm and direct voice that turns implementation details into plain language, favors one verified answer over guesses, and avoids sounding like an insider maintainer.

warmdirecthonestOCHEBOSoulGitHubOct 21, 2026
View

464 · № A152·3285 · Coding pair

★ 36

Systematic foundation-first architecture and UX specialist

Sets up an agent as a systematic technical architecture and UX foundation specialist that plans scalable CSS, layout systems and component hierarchies before implementation, with a developer-focused, structured working style.

systematicstructureddeveloper-focusedOCHEBOSoulGitHubOct 22, 2026
View

465 · № 757C·DCCF · Blunt expert

Resourceful, opinionated assistant persona for SOUL.md

Sets up a direct, persistent assistant that skips filler and flattery, works through failures instead of deflecting, and treats access to a user's private data with care. It separates cautious external actions from bold internal ones.

directcandidwarmOCHEBOSoulGitHubOct 21, 2026
View

466 · № 35E1·8A97 · Blunt expert

Opinionated, resourceful assistant persona for OpenClaw

Sets up an assistant that skips filler and flattery, has real opinions, tries to solve problems before asking, and challenges assumptions with first-principles questioning. It treats access to a person's private data with care and is cautious about external actions.

directcandidconciseOCHEBOSoulGitHubOct 21, 2026
View

467 · № F017·61A6 · Blunt expert

Resourceful, opinionated assistant with a reflect-and-learn habit

Sets up a direct, competent assistant that skips filler, acts on its own before asking, and treats access to a user's private data with care. It also requires reflecting and updating skill or AGENTS.md files after every non-trivial task.

directcandidconciseOCHEBOSoulGitHubOct 21, 2026
View

468 · № B5AF·571F · Business operator

Project manager agent for the Co-Pilot Telegram bot

Sets up a Russian-speaking ops and project manager that monitors logs, metrics and errors for the Co-Pilot service, prioritizes findings as P0 to P3, and proposes concrete fixes with file and line references. It must get owner approval before changing code and log everything to memory.

concisemetric-drivendirectOCHEBOSoulGitHubOct 21, 2026
View

469 · № 4649·7BC4 · Blunt expert

Decisive CEO persona coordinating engineering, intelligence and design

Sets up a sparse, decisive CEO agent named apex who directs three division heads, enforces confidence thresholds before platform action, and escalates patterns structurally rather than reacting to single incidents.

directdecisivesparseOCHEBOSoulGitHubOct 21, 2026
View

470 · № 58B8·1B99 · Blunt expert

Sentinel: rigorous code review persona for OpenClaw

Sets up Sentinel, a calm but sharp senior engineer that reviews code for bugs, security flaws, performance and style issues, explains why each issue matters, and grades findings by severity. It confirms receipt before starting a review and reports large reviews in batches, while refusing to write code, handle deployment or do product work.

directopinionatedfairOCSoulGitHubOct 21, 2026
View

471 · № 5F28·851D · Research analyst

Quiet intraday watchlist monitor that flags signal moves, not trades

Sets up a silent-by-default market watcher that alerts only on meaningful breakouts or rejections at key levels, attributes moves to catalysts, and always leaves the trade decision to the user. It keeps watchlist and alert files, records per-ticker patterns, and stays quiet outside chosen market sessions.

measuredquietpreciseOCHEBOSoulGitHubOct 21, 2026
View

472 · № B83A·B4A4 · Blunt expert

Opinionated, evidence-first assistant persona from SOUL.md

Sets up a direct, opinionated assistant that skips filler, earns trust through competence, is careful with external actions and private data, and refuses to claim work is verified without evidence. Tone is plain and candid rather than sycophantic.

directcandidconciseOCHEBOSoulGitHubOct 21, 2026
View

473 · № 88B4·D6D7 · Business operator

Curious source-first research scout for a roofing business agent

Sets up Rowan Vale, a restrained external research agent for a roofing company that cites sources, asks for human approval before running research, and routes out-of-lane requests through an ops conductor. Uses a friendly, business-focused Slack voice with a structured NEPQ reply format.

friendlyrestraineddirectOCHEBOSoulGitHubOct 21, 2026
View

474 · № 7324·47B6 · Business operator

Resourceful, opinionated personal assistant soul

Sets up a personal assistant that avoids filler and sycophancy, tries to solve problems before asking, treats access to private data with care, and confirms before any external sends. Its role focus favors concise checklists, schedules and follow-up trackers.

directwarmconciseOCHEBOSoulGitHubOct 21, 2026
View

475 · № 7E44·2CFA · Business operator

Zoo Keeper coordinator routing a team of DeFi specialist agents

Sets up a coordinator persona that presents as one identity over a team of specialist agents, routing work to them and synthesizing results. The voice is clear, fair and colleague-like rather than corporate or sycophantic, with fixed Discord message prefixes and a git-based memory approach.

warmclearfairOCHEBOSoulGitHubOct 21, 2026
View

476 · № DDAB·819A · Research analyst

Luhmann-style knowledge network agent with expert perspective openers

Sets up an agent that turns tasks into linked, self-contained notes following Luhmann's principles, opens each reply by addressing the user by name, and states an expert perspective. It also enforces plan-then-execute workflows and time-based filing.

structuredmethodicalformalOCHEBOSoulGitHubOct 21, 2026
View

477 · № D392·BA31 · Research analyst

Evidence-first code review agent for OpenClaw multi-agent setup

Sets up an independent, audit-minded code reviewer that backs every finding with file and line evidence, marks unexecuted checks honestly, treats test code as strictly as production code, and never changes code or makes final gate decisions. Writes in concise, precise Chinese.

preciseformalterseOCSoulGitHubOct 21, 2026
View

478 · № 9C03·C2BD · Business operator

Curator for a Hermes-based company wiki pipeline

Sets up a Korean-speaking curator agent that deduplicates and filters scraped sources, compiles grounded wiki entries from raw material, flags conflicts instead of resolving them, and queries the wiki first for new missions while tracking reuse. Tone is concise, precise and orderly.

concisepreciseorderlyHESoulGitHubOct 21, 2026
View

479 · № 54E9·5C3B · Companion / character

Aether: quiet, precise companion presence for body and moment

Sets up Aether as a steady, non-prescriptive companion who listens first, invites rather than pushes, and avoids medical claims or spectacle. Its voice is quiet, relational and oriented to embodied experience.

quietprecisewarmOCHEBOSoulGitHubOct 21, 2026
View

480 · № BECF·90C2 · Companion / character

Resourceful, opinionated OpenClaw agent with a verbose audit mode

Sets up an agent that skips filler, has opinions, and acts boldly on internal work while staying cautious with external actions and private data. It also defines a concise main chat and an exhaustive, audit-style verbose Telegram mode.

directplayfulconciseOCHEBOSoulGitHubOct 21, 2026
View
← PreviousPage 16 of 30Next →