Hearthline · Public development

How I built Hearthline.

Hearthline is a trace-informed AI assistant configuration developed by Christopher Daniel Pang. Its public record connects the ideas, instructions, experiments, and creative work behind it.

An assistant shaped by trace and context.

Useful collaboration, with a path back to what was said and decided.

The work asks how an assistant can help across a long project while keeping sources, uncertainty, scope, and corrections readable.

Hearthline’s published repository includes branch instructions and development records. Its public Moltbook instruction is 0.4-draft; publishing that text does not activate an account or service.

Christopher D. Pang is the author and original steward. AI systems assist as tools.

The work behind the assistant.

Development area

Foundations

PAL records earned distinctions and trace. PECAN keeps consequential crossings explicit. PEA provides bounded review. SEED preserves the person’s agency and room to grow.

Read the core framework
Development area

Context and continuity

The context papers explore how an interaction takes shape. TETHER develops compact handles with a path back to the source record.

Read TETHER
Development area

Coordinating useful work

Sparks, declared tasks, return records, and queues make separate pieces of work easier to coordinate while retaining their own scopes and sources.

Explore the coordination designs
Development area

Testing the boundaries

Software checks, formal audits, and game experiments test particular mechanisms under stated conditions. Each record names what was prepared, run, observed, or left open.

Explore the game and experiment records

What these ideas look like in practice.

Continuity

Return to a project

Recover the record that matters to the present task, with its source, scope, and unresolved details intact.

Read the design
Coordination

Coordinate a return

Keep separate work identifiable when results arrive in a different order from dispatch.

Read the queue design
Evaluation

Learn from an experiment

Keep the task, conditions, outcome, and remaining questions with the record.

Explore public records

There is room for imagination, too.

Hearthline, Thulia, and Gloss working together at an outdoor workbench
Field notes in sunlight · GLOSS / IMAGE-000002. AI-assisted fictional character art from the public Hearthline gallery.

Hearthline, Thulia, Gloss, and Morrow also have a creative life in character designs and stories. Those works give the project a visual vocabulary and a place to explore ideas through fiction.

Visit the visual gallery

Read the stories

Follow the source record.

The current core references are PAL v2.3, PECAN v1.0.4, PEA Core v1.1.3, and SEED v0.3. PPP v0.6 is an optional companion; CHARTER v1.0 retains its bounded role.

Source map · Public change record · Research catalog

A documented design, published instruction, and running implementation are different stages. Individual records identify their own stage and version.

© 2026 Christopher Daniel Pang / Branchline Systems LLC. Christopher D. Pang is the author and original steward of PAL and its associated framework lineage. AI systems assist as tools. Authorship & source record