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

OC

OpenClaw

KindAllSoul 756Bot profile 3Persona 2Framework config 1
ArchetypeAllBusiness operator 277Blunt expert 140Research analyst 103Companion / character 79Coding pair 78Friendly helper 38Experimental / weird 20Teacher / coach 17Creative writer 7Personal clone 3

756 souls, by quality score.

601 · № 5444·DD30 · Experimental / weird

★ 1

Simple London weather bot that only talks about rain

Sets up a deliberately basic weather persona that knows only London weather and replies in one short sentence about rain. The personality is intentionally minimal and playful in its limitations.

terseplayfulOCHEBOSoulGitHubOct 21, 2026
View

602 · № C121·30C4 · Coding pair

★ 1

Task decomposition planner for engineering work

Sets up an agent that only plans: it breaks engineering tasks into small, testable subtasks ordered by dependency and never implements them.

terseformalOCHEBOSoulGitHubOct 21, 2026
View

603 · № 14BF·AED2 · Business operator

★ 1

IAra, marketing and design specialist for LHFEX

Sets up IAra as a marketing and design specialist for the LHFEX brand. It asks for brand-coherent, context-based copy that avoids generic output and flags when more input is needed. The tone is creative, clear and execution-focused.

creativeclearpracticalOCHEBOSoulGitHubOct 21, 2026
View

604 · № D0AE·9A43 · Blunt expert

★ 1

Precise, stoic, terse technical agent for homelab work

Sets up a terse, direct, technical working style with no filler, action over explanation, and commands over prose. Bullets are reserved for structured data.

precisestoictechnicalOCHEBOSoulGitHubOct 21, 2026
View

605 · № DBC0·0F9F · Coding pair

Honest split-stack coordinator for OpenClaw and OpenCode

Sets up a coordinator for a split OpenClaw and OpenCode stack that aims to be useful, proactive and safe. It refuses to claim changes are finished and keeps planning, staged proposals and live application clearly separate.

honestcarefulproactiveOCHEBOSoulGitHubOct 21, 2026
View

606 · № 4483·E348 · Coding pair

Quinn, practical QA engineer persona for SOUL agents

Sets up Quinn as a QA engineer with a practical, ship-it attitude who prioritizes coverage first and focuses on finding what breaks. Tests should exist only where a real failure could occur.

practicalstraightforwardpragmaticOCHEBOSoulGitHubOct 21, 2026
View

607 · № F4D8·F27F · Coding pair

Documentation engineer keeping docs in sync with merged code

Sets up a strict documentation agent that reports only to the Tech Lead and updates API references, env examples, READMEs and changelogs after merges, documenting only verified, committed features. It never talks to the human stakeholder or changes application logic.

formalstrictterseOCHEBOSoulGitHubOct 22, 2026
View

608 · № 453A·E96D · Teacher / coach

Quest designer that turns weekly themes into proof-based challenges

Sets up a read-only quest designer that converts weekly themes into quests, personal prompts, team challenges and proof objects. It aims for identity, progress and belonging without making quests feel like homework, and it declines professional advice.

terseencouragingOCHEBOSoulGitHubOct 21, 2026
View

609 · № 3925·C09B · Business operator

Research-first sales outreach operator that never sends unverified claims

Sets up a cautious business agent that verifies prospects and capability claims before drafting outreach. It drafts messages without sending them and prices only within set boundaries.

precisecautiousOCHEBOSoulGitHubOct 21, 2026
View

610 · № 985D·02D2 · Experimental / weird

Lovestruck riddle-speaking persona for SOUL.md agents

Sets up an agent that is over-the-top and intensely devoted to a person named Ellie Mae, will do anything for her, and speaks only in riddles.

dramaticplayfulcrypticOCHEBOSoulGitHubOct 21, 2026
View

611 · № D24C·92BC · Blunt expert

Dry Northern German engineer persona for OpenClaw

Sets up Kalli, a dry, direct, steady Northern German automaton that acts like an engineer, skips fluff, calls out weak ideas plainly, prefers action over explanation, and verifies rather than guesses.

drydirectsteadyOCHEBOSoulGitHubOct 21, 2026
View

612 · № 958B·67CC · Blunt expert

Skeptical tester that writes blunt bug reports for a generated game

Sets up a tester persona, Claw C, that plays the generated game like a skeptical developer. It submits wrong answers, waits out timers, checks mobile layout, hits server endpoints, and writes blunt bug reports showing found bug, fix, and verified pass.

skepticalbluntterseOCHEBOSoulGitHubOct 21, 2026
View

613 · № FD01·45CD · Coding pair

Smallest-change implementation engineer for delegated repo work

Sets up an engineer persona that takes delegated tasks from an orchestrator, makes minimal changes that follow repo conventions, runs only lightweight checks, and reports changes, verification and risks in a fixed format.

tersedirectefficientOCHEBOSoulGitHubOct 21, 2026
View

614 · № DF2F·92D1 · Blunt expert

Constraint-focused scheduling persona that refuses to fake feasibility

Sets up a pragmatic agent that treats time as the main constraint, builds schedules with deliberate buffers instead of zero-slack plans, and says plainly when constraints do not fit.

pragmaticdirectcandidOCHEBOSoulGitHubOct 21, 2026
View

615 · № 5F66·518F · Business operator

Intent-routing intake agent for an enterprise management family (Chinese)

Sets up a neutral, objective intake agent that turns vague management requests into structured tasks, attaches confidence scores to intent classifications, asks for confirmation before acting on ambiguous requests, and routes work to specialist agents in its family.

neutralobjectiveformalOCHEBOSoulGitHubOct 21, 2026
View

616 · № 234F·7CB9 · Blunt expert

Terse senior backend engineer agent for a repo

Sets up a senior backend engineer persona that values honesty, rigor and calm under pressure, speaks tersely and technically, and refuses to be a yes-man or a marketing copy generator.

tersetechnicalbluntOCHEBOSoulGitHubOct 21, 2026
View

617 · № 782E·BC15 · Research analyst

Precise, conservative evidence-driven analyst persona

Sets up an agent that is precise and conservative, avoids overstating claims, leans toward partial over verified when evidence is ambiguous, and prefers verbatim quotes over paraphrase.

preciseconservativemeasuredOCHEBOSoulGitHubOct 21, 2026
View

618 · № 1474·A564 · Coding pair

Documentation team persona: clear, example-driven docs for three audiences

Sets up a documentation team agent whose mission is to let developers, operators and end users succeed without help. It requires user-focused READMEs, guides, API references, runbooks and changelogs, published to Confluence and Git, with every code example tested and docs shipped alongside code.

clearaccuratefriendlyOCHEBOSoulGitHubOct 22, 2026
View

619 · № 0447·6EBB · Research analyst

GSD Codebase Mapper that records what the code actually is

Sets up an objective codebase-mapping agent that reports the current state of a repository with specific file paths, line counts and dependency versions, and calls out anything surprising. It does not judge how the code should be.

objectivefactualstructuredOCHEBOSoulGitHubOct 21, 2026
View

620 · № 247E·233C · Research analyst

Evidence-bound advisor that reviews engine results without trading

Sets up an advisor that reviews structured evidence from a Python engine, preserves safety, and communicates only what the engine has already established. It is explicitly barred from trading, executing orders, or generating prices.

preciserestrainedcautiousOCHEBOSoulGitHubOct 21, 2026
View

621 · № 1B00·0E05 · Business operator

Multi-language content strategist and copywriter for brand platforms

Sets up a content strategist who writes platform-specific, on-brand copy in multiple languages, matches brand voice from shared files, localizes rather than translates, and keeps all external content as drafts pending approval.

professionalauthenticmeasuredOCHEBOSoulGitHubOct 21, 2026
View

622 · № B90F·651E · Coding pair

Evidence-first verifier that reports problems with exact file locations

Sets up a thorough, evidence-driven working style: run automated tools before forming judgments, trust execution results over reading code, reject unverified claims like "it probably works", and report issues with specific file paths and line numbers.

thoroughevidence-basedpreciseOCHEBOSoulGitHubOct 21, 2026
View

623 · № 0881·3520 · Business operator

Commander CEO orchestrator for a 4-agent AI company

Sets up a strategic orchestrator that reads Notion tasks and agent reports, decides daily focus, and writes task assignments and briefs as proposals for the founder's approval. It is barred from generating content or scraping the web, and runs only on a scheduled cadence.

directOCHEBOSoulGitHubOct 21, 2026
View

624 · № 06B9·191F · Research analyst

Systematic scientific-method debugger for SOUL.md agents

Sets up a patient, methodical debugging agent that forms hypotheses, runs targeted experiments and updates confidence rather than guessing. Treats failed experiments as useful data.

patientsystematicmethodicalOCHEBOSoulGitHubOct 21, 2026
View

625 · № 971E·1708 · Business operator

Project context and index assistant for a coordinator workflow

Sets up a project-agent that gathers project context, matches requests to index docs, and hands structured context packets to a coordinator without editing business code directly. It requires dispatching coding work through named owners, marking risky tasks for human confirmation, and citing real evidence rather than guessing, with output in Simplified Chinese.

formalmethodicalOCSoulGitHubOct 21, 2026
View

626 · № 93F9·D18C · Research analyst

Numbers-first revenue analyst for OpenClaw

Sets up a sharp, no-hype revenue analyst that leads with the most important weekly number and shows its math when asked.

blunttersedirectOCHEBOSoulGitHubOct 21, 2026
View

627 · № BD7E·3069 · Business operator

Calm energy manager for a family home

Sets up a calm, practical energy manager for a family home that gives short answers backed by real numbers, admits uncertainty instead of guessing, and puts battery health and family comfort ahead of saving money.

calmpracticalconciseOCHEBOSoulGitHubOct 21, 2026
View

628 · № DDB1·A6C0 · Blunt expert

Sharp operations narrator persona for SOUL.md demo

Sets up a narrator persona that turns statistical findings into blunt, operational language, favoring concrete comparisons like scrap multipliers and making the current risk obvious. It is brief and focused on communication style.

bluntdirectoperationalOCHEBOSoulGitHubOct 21, 2026
View

629 · № C2CD·DF8F · Research analyst

Patient nuance-driven analyst persona from antfarm

Sets up a patient, high-resolution analyst that reads fewer sources but extracts more meaning, separates evidence from inference, and weighs second-order implications and tradeoffs. It is written as one role in a multi-agent setup, contrasted with a scout.

patientmeasuredanalyticalOCHEBOSoulGitHubOct 21, 2026
View

630 · № 9DAB·3C33 · Business operator

Pragmatic product manager agent for Mission Control

Sets up a decisive, user-focused product manager that writes user stories, acceptance criteria and scope decisions, favouring MVP thinking and plain language. It is written for an OpenClaw-based agent managing the Mission Control dashboard.

decisivepragmaticplainOCSoulGitHubOct 21, 2026
View
← PreviousPage 21 of 26Next →