EDA Workflows¶

Context Teleport includes first-class support for Electronic Design Automation (EDA) projects. It can detect your project type, parse design artifacts into structured knowledge, and suggest relevant skills for your workflow.

Project Detection¶

The detect_eda_project() function scans a directory for markers that indicate the type of EDA project. Detection is automatic when you run EDA-related commands.

Supported Project Types¶

Detection also reads the PDK_ROOT environment variable as an additional marker and attempts to infer the PDK name from it.

Suggested Skills per Project Type¶

When a project type is detected, Context Teleport suggests relevant skills from the EDA skills pack:

EDA Artifact Parsers¶

Context Teleport ships six parsers that convert EDA tool outputs into structured knowledge entries.

Available Parsers¶

Auto-Detection¶

When you do not specify --type, the CLI tries each parser in turn and uses the first one that reports it can handle the file. This works well for unambiguous file formats (e.g., a .lib file is always Liberty, a config.json with meta.version is always LibreLane).

If auto-detection fails, specify the parser explicitly with --type.

Importing EDA Artifacts¶

Basic Import¶

# Auto-detect parser
context-teleport import eda /path/to/config.json

# Force a specific parser
context-teleport import eda /path/to/report.txt --type magic-drc

# Preview what would be imported
context-teleport import eda /path/to/metrics.json --dry-run

Key Naming and Re-Import¶

Each parser generates a deterministic key based on the design name extracted from the artifact. For example, importing a LibreLane config for design inverter produces the key librelane-config-inverter.

Re-importing the same artifact overwrites the existing entry with updated content. This is intentional: as your design iterates through synthesis, place-and-route, and signoff, you import updated artifacts and the knowledge store reflects the latest state.

JSON Output¶

All import commands support --format json for scripting:

context-teleport import eda /path/to/config.json --format json

{
  "items": [
    {"type": "knowledge", "key": "librelane-config-inverter", "source": "config.json"}
  ],
  "imported": 1,
  "dry_run": false,
  "importer": "librelane-config"
}

Example: LibreLane Design with IHP SG13G2¶

This example walks through importing artifacts for a LibreLane digital design targeting the IHP SG13G2 PDK.

1. Initialize the Context Store¶

cd /path/to/my-librelane-project
context-teleport init my-inverter

2. Import the LibreLane Configuration¶

context-teleport import eda config.json
# Auto-detected: librelane-config
# Imported 1 item(s) via librelane-config
#   librelane-config-inverter <- config.json

The imported knowledge entry contains a structured summary of the configuration: design name, PDK, clock period, die area, pin placements, and key flow parameters.

3. Run DRC and Import Results¶

After running Magic DRC through the LibreLane flow:

context-teleport import eda results/signoff/inverter.drc
# Imported 1 item(s) via magic-drc
#   drc-summary-inverter <- inverter.drc

4. Run LVS and Import Results¶

context-teleport import eda results/signoff/inverter.lvs
# Imported 1 item(s) via netgen-lvs
#   lvs-summary-inverter <- inverter.lvs

5. Check Imported Knowledge¶

context-teleport knowledge list
# key                          scope    updated
# librelane-config-inverter    public   2025-06-15
# drc-summary-inverter         public   2025-06-15
# lvs-summary-inverter         public   2025-06-15

Now any AI agent connected via MCP has full visibility into the design configuration, DRC status, and LVS results. When the agent encounters a DRC violation, it can cross-reference the config to understand pin placements, die area constraints, and PDK-specific rules.

6. Re-Import After Iteration¶

After fixing DRC violations and re-running signoff:

context-teleport import eda results/signoff/inverter.drc
# Imported 1 item(s) via magic-drc
#   drc-summary-inverter <- inverter.drc

The entry is overwritten with the updated results. No duplicate keys accumulate.

Generic Artifact Import¶

The import eda command is a convenient shortcut for EDA-specific files, but Context Teleport also provides a generic import artifacts command that uses the full plugin registry. This registry includes all built-in EDA parsers plus any third-party importers installed via the ctx.importers entry-point group.

# Equivalent to "import eda" but uses the plugin registry
context-teleport import artifacts /path/to/config.json

# List all registered importers (built-in + plugins)
context-teleport import artifacts --list

If you have domain-specific artifact formats beyond EDA, you can write a plugin importer and it will be available through this command. See Adding Domain Importers for details.

EDA Skills Pack¶

Context Teleport provides a separate git repository (eda-skills-pack) containing pre-built SKILL.md files for common EDA tasks. These skills teach AI agents how to:

• Configure LibreLane flows
• Set up power delivery networks
• Debug DRC violations
• Debug LVS mismatches
• Analyze timing reports
• Run xschem simulations
• Characterize devices
• Port designs across PDKs

To use the skills pack, clone it and import the skills into your project:

# Clone the skills pack
git clone https://github.com/your-org/eda-skills-pack.git /tmp/eda-skills-pack

# Import individual skills
context-teleport skill add debug-drc --file /tmp/eda-skills-pack/skills/debug-drc/SKILL.md

Note: A future version will support context-teleport skill install <pack-url> for direct installation from a repository URL.